,Clean_Title,Clean_Text,Clean_Summary
0,FearNet: Brain-Inspired Model for Incremental Learning,"Incremental class learning involves sequentially learning classes in bursts of examples from the same class.This violates the assumptions that underlie methods for training standard deep neural networks, and will cause them to suffer from catastrophic forgetting.Arguably, the best method for incremental class learning is iCaRL, but it requires storing training examples for each class, making it challenging to scale.Here, we propose FearNet for incremental class learning.FearNet is a generative model that does not store previous examples, making it memory efficient.FearNet uses a brain-inspired dual-memory system in which new memories are consolidated from a network for recent memories inspired by the mammalian hippocampal complex to a network for long-term storage inspired by medial prefrontal cortex.Memory consolidation is inspired by mechanisms that occur during sleep.FearNet also uses a module inspired by the basolateral amygdala for determining which memory system to use for recall. FearNet achieves state-of-the-art performance at incremental class learning on image and audio classification benchmarks.","FearNet is a memory efficient neural-network, inspired by memory formation in the mammalian brain, that is capable of incremental class learning without catastrophic forgetting.This paper presents a novel solution to an incremental classification problem based on a dual memory system. "
1,Improving Sentence Representations with Multi-view Frameworks,"Multi-view learning can provide self-supervision when different views are available of the same data.Distributional hypothesis provides another form of useful self-supervision from adjacent sentences which are plentiful in large unlabelled corpora.Motivated by the asymmetry in the two hemispheres of the human brain as well as the observation that different learning architectures tend to emphasise different aspects of sentence meaning, we present two multi-view frameworks for learning sentence representations in an unsupervised fashion.One framework uses a generative objective and the other a discriminative one.In both frameworks, the final representation is an ensemble of two views, in which, one view encodes the input sentence with a Recurrent Neural Network, and the other view encodes it with a simple linear model.We show that, after learning, the vectors produced by our multi-view frameworks provide improved representations over their single-view learnt counterparts, and the combination of different views gives representational improvement over each view and demonstrates solid transferability on standard downstream tasks.","Multi-view learning improves unsupervised sentence representation learningApproach uses different, complementary encoders of the input sentence and consensus maximization.The paper presents a multi-view framework for improving sentence representation in NLP tasks using generative and discriminative objective architectures.This paper shows that multi-view frameworks are more effective than using individual encoders for learning sentence representations."
2,Learning objects from pixels,"We show how discrete objects can be learnt in an unsupervised fashion from pixels, and how to perform reinforcement learning using this object representation.More precisely, we construct a differentiable mapping from an image to a discrete tabular list of objects, where each object consists of a differentiable position, feature vector, and scalar presence value that allows the representation to be learnt using an attention mechanism.Applying this mapping to Atari games, together with an interaction net-style architecture for calculating quantities from objects, we construct agents that can play Atari games using objects learnt in an unsupervised fashion.During training, many natural objects emerge, such as the ball and paddles in Pong, and the submarine and fish in Seaquest.This gives the first reinforcement learning agent for Atari with an interpretable object representation, and opens the avenue for agents that can conduct object-based exploration and generalization.","We show how discrete objects can be learnt in an unsupervised fashion from pixels, and how to perform reinforcement learning using this object representation.A method for learning object representations from pixels for doing reinforcement learning. The paper proposes a neural architecture to map video streams to a discrete collection of objects, without human annotations, using an unsupervised pixel reconstruction loss. "
3,Learning what and where to attend,"Most recent gains in visual recognition have originated from the inclusion of attention mechanisms in deep convolutional networks.Because these networks are optimized for object recognition, they learn where to attend using only a weak form of supervision derived from image class labels.Here, we demonstrate the benefit of using stronger supervisory signals by teaching DCNs to attend to image regions that humans deem important for object recognition.We first describe a large-scale online experiment used to supplement ImageNet with nearly half a million human-derived ""top-down"" attention maps.Using human psychophysics, we confirm that the identified top-down features from ClickMe are more diagnostic than ""bottom-up"" saliency features for rapid image categorization.As a proof of concept, we extend a state-of-the-art attention network and demonstrate that adding ClickMe supervision significantly improves its accuracy and yields visual features that are more interpretable and more similar to those used by human observers.","A large-scale dataset for training attention models for object recognition leads to more accurate, interpretable, and human-like object recognition.Argues recent gains in visual recognition stem from using visual attention mechanisms in deep convolutional networks, which learn where to focus through a weak form of supervision based on image class labels.Presents a new take on attention in which a large attention dataset is collected and used to train a NN in a supervised manner to exploit self-reported human attention.This paper proposes a new approach to use more informative signals, specifically, regions humans deem important on images, to improve deep convolutional neural networks."
4,EFFICIENT TWO-STEP ADVERSARIAL DEFENSE FOR DEEP NEURAL NETWORKS,"In recent years, deep neural networks have demonstrated outstanding performancein many machine learning tasks.However, researchers have discovered that thesestate-of-the-art models are vulnerable to adversarial examples: legitimate examples added by small perturbations which are unnoticeable to human eyes.Adversarial training, which augments the training data with adversarial examples duringthe training process, is a well known defense to improve the robustness of themodel against adversarial attacks. However, this robustness is only effective tothe same attack method used for adversarial training. Madry et al. suggest that effectiveness of iterative multi-step adversarial attacks and particularlythat projected gradient descent may be considered the universal first order adversary and applying the adversarial training with PGD implies resistanceagainst many other first order attacks. However, the computational cost of theadversarial training with PGD and other multi-step adversarial examples is muchhigher than that of the adversarial training with other simpler attack techniques.In this paper, we show how strong adversarial examples can be generated only ata cost similar to that of two runs of the fast gradient sign method, allowing defense against adversarial attacks with a robustness level comparable to thatof the adversarial training with multi-step adversarial examples. We empiricallydemonstrate the effectiveness of the proposed two-step defense approach againstdifferent attack methods and its improvements over existing defense strategies.","We proposed a time-efficient defense method against one-step and iterative adversarial attacks.Propose a novel, computationaly efficient method named e2SAD which generates sets of two training adversarial samples for each clean training sample.The paper introduces a two-step adversarial defense method, to generate two adversarial examples per clean sample and include them in the actual training loop to achieve robustness and claiming it can outperform more expensive iterative methods.The paper presents a 2-step approach to generate strong adversarial examples at a far lesser cost as compared to recent iterative multi-step adversarial attacks."
5,Character Level Based Detection of DGA Domain Names,"Recently several different deep learning architectures have been proposed that take a string of characters as the raw input signal and automatically derive features for text classification.Little studies are available that compare the effectiveness of these approaches for character based text classification with each other.In this paper we perform such an empirical comparison for the important cybersecurity problem of DGA detection: classifying domain names as either benign vs. produced by malware.Training and evaluating on a dataset with 2M domain names shows that there is surprisingly little difference between various convolutional neural network and recurrent neural network based architectures in terms of accuracy, prompting a preference for the simpler architectures, since they are faster to train and less prone to overfitting.","A comparison of five deep neural network architectures for detection of malicious domain names shows surprisingly little difference.Authors propose using five deep architectures for the cybersecurity task of domain generation algorithm detection. Applies several NN architectures to classify url's between begign and malware related URLs."", 'This paper proposes to automatically recognize domain names as malicious or benign by deep networks trained to directly classify the character sequence as such."
6,Mitigating Bias in Natural Language Inference Using Adversarial Learning,"Recognizing the relationship between two texts is an important aspect of natural language understanding, and a variety of neural network models have been proposed for solving NLU tasks.Unfortunately, recent work showed that the datasets these models are trained on often contain biases that allow models to achieve non-trivial performance without possibly learning the relationship between the two texts.We propose a framework for building robust models by using adversarial learning to encourage models to learn latent, bias-free representations.We test our approach in a Natural Language Inference scenario, and show that our adversarially-trained models learn robust representations that ignore known dataset-specific biases.Our experiments demonstrate that our models are more robust to new NLI datasets.",Adversarial learning methods encourage NLI models to ignore dataset-specific biases and help models transfer across datasets.The paper proposes an adversarial setup to mitigate annotation artifacts in natural language inference dataThis paper presents a method for removing bias of a textual entailment model through an adversarial training objective. 
7,RotatE: Knowledge Graph Embedding by Relational Rotation in Complex Space,"We study the problem of learning representations of entities and relations in knowledge graphs for predicting missing links.The success of such a task heavily relies on the ability of modeling and inferring the patterns of the relations.In this paper, we present a new approach for knowledge graph embedding called RotatE, which is able to model and infer various relation patterns including: symmetry/antisymmetry, inversion, and composition.Specifically, the RotatE model defines each relation as a rotation from the source entity to the target entity in the complex vector space.In addition, we propose a novel self-adversarial negative sampling technique for efficiently and effectively training the RotatE model.Experimental results on multiple benchmark knowledge graphs show that the proposed RotatE model is not only scalable, but also able to infer and model various relation patterns and significantly outperform existing state-of-the-art models for link prediction.","A new state-of-the-art approach for knowledge graph embedding.Presents a neural link prediction scoring function that can infer symmetry, anti-symmetry, inversion and composition patterns of relations in a knowledge base.This paper proposes an approach to knowledge graph embedding by modeling relations as rotations in the complex vector space.Proposes a method for graph embedding to be used for link prediction"
8,Meta-learning with differentiable closed-form solvers,"Adapting deep networks to new concepts from a few examples is challenging, due to the high computational requirements of standard fine-tuning procedures.Most work on few-shot learning has thus focused on simple learning techniques for adaptation, such as nearest neighbours or gradient descent.Nonetheless, the machine learning literature contains a wealth of methods that learn non-deep models very efficiently.In this paper, we propose to use these fast convergent methods as the main adaptation mechanism for few-shot learning.The main idea is to teach a deep network to use standard machine learning tools, such as ridge regression, as part of its own internal model, enabling it to quickly adapt to novel data.This requires back-propagating errors through the solver steps.While normally the cost of the matrix operations involved in such a process would be significant, by using the Woodbury identity we can make the small number of examples work to our advantage.We propose both closed-form and iterative solvers, based on ridge regression and logistic regression components.Our methods constitute a simple and novel approach to the problem of few-shot learning and achieve performance competitive with or superior to the state of the art on three benchmarks.","We propose a meta-learning approach for few-shot classification that achieves strong performance at high-speed by back-propagating through the solution of fast solvers, such as ridge regression or logistic regression.The paper proposes an algorithm for meta-learning which amounts to fixing the features (ie all hidden layers of a deep NN), and treating each task as having its own final layer which could be a ridge regression or a logistic regression.This paper proposes a meta-learning approach for the problem of few-shot classification, they use a method based on parametrizing the learner for each task by a closed-form solver."
9,Active Learning with Partial Feedback,"While many active learning papers assume that the learner can simply ask for a label and receive it, real annotation often presents a mismatch between the form of a label, and the form of an annotation.To annotate examples corpora for multiclass classification, we might need to ask multiple yes/no questions, exploiting a label hierarchy if one is available.To address this more realistic setting, we propose active learning with partial feedback, where the learner must actively choose both which example to label and which binary question to ask.At each step, the learner selects an example, asking if it belongs to a chosen class.Each answer eliminates some classes, leaving the learner with a partial label.The learner may then either ask more questions about the same example or move on immediately, leaving the first example partially labeled.Active learning with partial labels requires a sampling strategy to choose pairs, and learning from partial labels between rounds.Experiments on Tiny ImageNet demonstrate that our most effective method improves 26% in top-1 classification accuracy compared to i.i.d. baselines and standard active learners given 30% of the annotation budget that would be required to annotate the dataset.Moreover, ALPF-learners fully annotate TinyImageNet at 42% lower cost.Surprisingly, we observe that accounting for per-example annotation costs can alter the conventional wisdom that active learners should solicit labels for hard examples.","We provide a new perspective on training a machine learning model from scratch in hierarchical label setting, i.e. thinking of it as two-way communication between human and algorithms, and study how we can both measure and improve the efficiency. Introduces a new Active Learning setting where the oracle offers a partial or weak label instead of querying for a particular example's label, leading to a simpler retrieval of information."", 'This paper proposes a method of active learning with partial feedback that outperforms existing baselines under a limited budget.The paper considers a multiclass classification problem in which labels are grouped in a given number M of subsets, which contain all individual labels as singletons."
10,Learning Embeddings into Entropic Wasserstein Spaces,"Despite their prevalence, Euclidean embeddings of data are fundamentally limited in their ability to capture latent semantic structures, which need not conform to Euclidean spatial assumptions.Here we consider an alternative, which embeds data as discrete probability distributions in a Wasserstein space, endowed with an optimal transport metric.Wasserstein spaces are much larger and more flexible than Euclidean spaces, in that they can successfully embed a wider variety of metric structures.We propose to exploit this flexibility by learning an embedding that captures the semantic information in the Wasserstein distance between embedded distributions.We examine empirically the representational capacity of such learned Wasserstein embeddings, showing that they can embed a wide variety of complex metric structures with smaller distortion than an equivalent Euclidean embedding.We also investigate an application to word embedding, demonstrating a unique advantage of Wasserstein embeddings: we can directly visualize the high-dimensional embedding, as it is a probability distribution on a low-dimensional space.This obviates the need for dimensionality reduction techniques such as t-SNE for visualization.","We show that Wasserstein spaces are good targets for embedding data with complex semantic structure.Learns embeddings in a discrete space of probability distributions, using a minimized, regularised version of Wasserstein distances.The paper describes a new embedding method that embeds data to the space of probability measures endowed with the Wasserstein distance. The paper proposes embedding the data into low-dimensional Wasserstein spaces, which can capture the underlying structure of the data more accurately."
11,Deep Continuous Clustering,"Clustering high-dimensional datasets is hard because interpoint distances become less informative in high-dimensional spaces.We present a clustering algorithm that performs nonlinear dimensionality reduction and clustering jointly.The data is embedded into a lower-dimensional space by a deep autoencoder.The autoencoder is optimized as part of the clustering process.The resulting network produces clustered data.The presented approach does not rely on prior knowledge of the number of ground-truth clusters.Joint nonlinear dimensionality reduction and clustering are formulated as optimization of a global continuous objective.We thus avoid discrete reconfigurations of the objective that characterize prior clustering algorithms.Experiments on datasets from multiple domains demonstrate that the presented algorithm outperforms state-of-the-art clustering schemes, including recent methods that use deep networks.","A clustering algorithm that performs joint nonlinear dimensionality reduction and clustering by optimizing a global continuous objective.Presents a clustering algorithm by jointly solving deep autoencoder and clustering as a global continuous objective, showing better results than state-of-the-art clustering schemas.Deep Continuous Clustering is a clustering method that integrates the autoencoder objective with the clustering objective then train using SGD."
12,Spatial-Winograd Pruning Enabling Sparse Winograd Convolution,"Deep convolutional neural networks are deployed in various applications but demand immense computational requirements.Pruning techniques and Winograd convolution are two typical methods to reduce the CNN computation.However, they cannot be directly combined because Winograd transformation fills in the sparsity resulting from pruning.Li et al. propose sparse Winograd convolution in which weights are directly pruned in the Winograd domain, but this technique is not very practical because Winograd-domain retraining requires low learning rates and hence significantly longer training time.Besides, Liu et al. move the ReLU function into the Winograd domain, which can help increase the weight sparsity but requires changes in the network structure.To achieve a high Winograd-domain weight sparsity without changing network structures, we propose a new pruning method, spatial-Winograd pruning.As the first step, spatial-domain weights are pruned in a structured way, which efficiently transfers the spatial-domain sparsity into the Winograd domain and avoids Winograd-domain retraining.For the next step, we also perform pruning and retraining directly in the Winograd domain but propose to use an importance factor matrix to adjust weight importance and weight gradients.This adjustment makes it possible to effectively retrain the pruned Winograd-domain network without changing the network structure.For the three models on the datasets of CIFAR-10, CIFAR-100, and ImageNet, our proposed method can achieve the Winograd-domain sparsities of 63%, 50%, and 74%, respectively.","To accelerate the computation of convolutional neural networks, we propose a new two-step pruning technique which achieves a higher Winograd-domain weight sparsity without changing the network structure.Proposes a spatial-Winograd pruning framework which allows pruned weight from the spatial domain to be kept in the Winograd domain and improves the sparsity of the Winograd domain.Proposes two techniques for pruning convolutional layers which use the Winograd algorithm"
13,Probabilistic Federated Neural Matching,"In federated learning problems, data is scattered across different servers and exchanging or pooling it is often impractical or prohibited.We develop a Bayesian nonparametric framework for federated learning with neural networks.Each data server is assumed to train local neural network weights, which are modeled through our framework.We then develop an inference approach that allows us to synthesize a more expressive global network without additional supervision or data pooling.We then demonstrate the efficacy of our approach on federated learning problems simulated from two popular image classification datasets.","We propose a Bayesian nonparametric model for federated learning with neural networks.Uses beta process to do federated neural matching.The paper considers federate learning of neural networks, where data is distributed on multiple machines and the allocation of data points is potentially inhomogenous and unbalanced."
14,Generalizing Hamiltonian Monte Carlo with Neural Networks,"We present a general-purpose method to train Markov chain Monte Carlo kernels, parameterized by deep neural networks, that converge and mix quickly to their target distribution.Our method generalizes Hamiltonian Monte Carlo and is trained to maximize expected squared jumped distance, a proxy for mixing speed.We demonstrate large empirical gains on a collection of simple but challenging distributions, for instance achieving a 106x improvement in effective sample size in one case, and mixing when standard HMC makes no measurable progress in a second.Finally, we show quantitative and qualitative gains on a real-world task: latent-variable generative modeling.Python source code will be open-sourced with the camera-ready paper.","General method to train expressive MCMC kernels parameterized with deep neural networks. Given a target distribution p, our method provides a fast-mixing sampler, able to efficiently explore the state space.Proposes a generalized HMC by modifying the leapfrog integrator using neural networks to make the sampler to converge and mix quickly. "
15,Rigorous Agent Evaluation: An Adversarial Approach to Uncover Catastrophic Failures,"This paper addresses the problem of evaluating learning systems in safety critical domains such as autonomous driving, where failures can have catastrophic consequences.We focus on two problems: searching for scenarios when learned agents fail and assessing their probability of failure.The standard method for agent evaluation in reinforcement learning, Vanilla Monte Carlo, can miss failures entirely, leading to the deployment of unsafe agents.We demonstrate this is an issue for current agents, where even matching the compute used for training is sometimes insufficient for evaluation.To address this shortcoming, we draw upon the rare event probability estimation literature and propose an adversarial evaluation approach.Our approach focuses evaluation on adversarially chosen situations, while still providing unbiased estimates of failure probabilities.The key difficulty is in identifying these adversarial situations -- since failures are rare there is little signal to drive optimization.To solve this we propose a continuation approach that learns failure modes in related but less robust agents.Our approach also allows reuse of data already collected for training the agent.We demonstrate the efficacy of adversarial evaluation on two standard domains: humanoid control and simulated driving.Experimental results show that our methods can find catastrophic failures and estimate failures rates of agents multiple orders of magnitude faster than standard evaluation schemes, in minutes to hours rather than days.","We show that rare but catastrophic failures may be missed entirely by random testing, which poses issues for safe deployment. Our proposed approach for adversarial testing fixes this.Proposes a method which learns a failure probability predictor for a learned agent, leading to predictions of which initial states cause a system to fail.This paper proposes an importance sampling approach to sampling failure cases for RL algorithms based on a function learned via a neural network on failures that occur during agent trainingThis paper proposed an adversarial approach to identifying catastrophic failure cases in reinforcement learning."
16,Lagging Inference Networks and Posterior Collapse in Variational Autoencoders,"The variational autoencoder is a popular combination of deep latent variable model and accompanying variational learning technique.By using a neural inference network to approximate the model's posterior on latent variables, VAEs efficiently parameterize a lower bound on marginal data likelihood that can be optimized directly via gradient methods.In practice, however, VAE training often results in a degenerate local optimum known as ""posterior collapse"" where the model learns to ignore the latent variable and the approximate posterior mimics the prior.In this paper, we investigate posterior collapse from the perspective of training dynamics.We find that during the initial stages of training the inference network fails to approximate the model's true posterior, which is a moving target.As a result, the model is encouraged to ignore the latent encoding and posterior collapse occurs.Based on this observation, we propose an extremely simple modification to VAE training to reduce inference lag: depending on the model's current mutual information between latent variable and observation, we aggressively optimize the inference network before performing each model update.Despite introducing neither new model components nor significant complexity over basic VAE, our approach is able to avoid the problem of collapse that has plagued a large amount of previous work.Empirically, our approach outperforms strong autoregressive baselines on text and image benchmarks in terms of held-out likelihood, and is competitive with more complex techniques for avoiding collapse while being substantially faster.","To address posterior collapse in VAEs, we propose a novel yet simple training procedure that aggressively optimizes inference network with more updates. This new training procedure mitigates posterior collapse and leads to a better VAE model. Looks into the phenomenon of posterior collapse, showing that increased training of the inference network can reduce the problem and lead to better optima.Authors propose changing the training procedure of VAEs only as a solution to posterior collapse, leaving the model and objective untouched."
17,Generative Discovery of Relational Medical Entity Pairs,"Online healthcare services can provide the general public with ubiquitous access to medical knowledge and reduce the information access cost for both individuals and societies.To promote these benefits, it is desired to effectively expand the scale of high-quality yet novel relational medical entity pairs that embody rich medical knowledge in a structured form.To fulfill this goal, we introduce a generative model called Conditional Relationship Variational Autoencoder, which can discover meaningful and novel relational medical entity pairs without the requirement of additional external knowledge.Rather than discriminatively identifying the relationship between two given medical entities in a free-text corpus, we directly model and understand medical relationships from diversely expressed medical entity pairs.The proposed model introduces the generative modeling capacity of variational autoencoder to entity pairs, and has the ability to discover new relational medical entity pairs solely based on the existing entity pairs.Beside entity pairs, relationship-enhanced entity representations are obtained as another appealing benefit of the proposed method.Both quantitative and qualitative evaluations on real-world medical datasets demonstrate the effectiveness of the proposed method in generating relational medical entity pairs that are meaningful and novel.","Generatively discover meaningful, novel entity pairs with a certain medical relationship by purely learning from the existing meaningful entity pairs, without the requirement of additional text corpus for discriminative extraction.Presents a variational autoencoder for generating entity pairs given a relation in a medical setting.In the medical context, this paper describes the classic problem of ""knowledge base completion"" from structured data only."
18,Diagnosing and Enhancing VAE Models,"Although variational autoencoders represent a widely influential deep generative model, many aspects of the underlying energy function remain poorly understood. In particular, it is commonly believed that Gaussian encoder/decoder assumptions reduce the effectiveness of VAEs in generating realistic samples. In this regard, we rigorously analyze the VAE objective, differentiating situations where this belief is and is not actually true. We then leverage the corresponding insights to develop a simple VAE enhancement that requires no additional hyperparameters or sensitive tuning. Quantitatively, this proposal produces crisp samples and stable FID scores that are actually competitive with a variety of GAN models, all while retaining desirable attributes of the original VAE architecture. The code for our model is available at .","We closely analyze the VAE objective function and draw novel conclusions that lead to simple enhancements.Proposes a two-stage VAE method to generate high-quality samples and avoid blurriness.This paper analyzes the Gaussian VAEs.The paper provides a number of theoretical results on ""vanilla"" Gaussian Variational Auto-Encoders, which are then used to build a new algorithm called ""2 stage VAEs""."
19,Hyperparameter optimization: a spectral approach,"We give a simple, fast algorithm for hyperparameter optimization inspired by techniques from the analysis of Boolean functions. We focus on the high-dimensional regime where the canonical example is training a neural network with a large number of hyperparameters.The algorithm an iterative application of compressed sensing techniques for orthogonal polynomials requires only uniform sampling of the hyperparameters and is thus easily parallelizable.Experiments for training deep neural networks on Cifar-10 show that compared to state-of-the-art tools, our algorithm finds significantly improved solutions, in some cases better than what is attainable by hand-tuning. In terms of overall running time, we are at least an order of magnitude faster than Hyperband and Bayesian Optimization. We also outperform Random Search. Our method is inspired by provably-efficient algorithms for learning decision trees using the discrete Fourier transform. We obtain improved sample-complexty bounds for learning decision trees while matching state-of-the-art bounds on running time.","A hyperparameter tuning algorithm using discrete Fourier analysis and compressed sensingInvestigates problem of optimizing hyperparameters under the assumption that the unknown function can be approximated, showing that the approximate minimization can be performed over the boolean hypercube.The paper explores hyperparameter optimization by assuming structure in the unknown function mapping hyperparameters to classification accuracy"
20,Learning Latent Permutations with Gumbel-Sinkhorn Networks,"Permutations and matchings are core building blocks in a variety of latent variable models, as they allow us to align, canonicalize, and sort data.Learning in such models is difficult, however, because exact marginalization over these combinatorial objects is intractable.In response, this paper introduces a collection of new methods for end-to-end learning in such models that approximate discrete maximum-weight matching using the continuous Sinkhorn operator. Sinkhorn iteration is attractive because it functions as a simple, easy-to-implement analog of the softmax operator.With this, we can define the Gumbel-Sinkhorn method, an extension of the Gumbel-Softmax method to distributions over latent matchings.We demonstrate the effectiveness of our method by outperforming competitive baselines on a range of qualitatively different tasks: sorting numbers, solving jigsaw puzzles, and identifying neural signals in worms.","A new method for gradient-descent inference of permutations, with applications to latent matching inference and supervised learning of permutations with neural networksThe paper utilizes finite approximation of the Sinkhorn operator to describe how one can construct a neural network for learning from permutation valued training data. The paper proposes a new method that approximates the discrete max-weight for learning latent permutations"
21,Mixed Precision Quantization of ConvNets via Differentiable Neural Architecture Search,"Recent work in network quantization has substantially reduced the time and space complexity of neural network inference, enabling their deployment on embedded and mobile devices with limited computational and memory resources.However, existing quantization methods often represent all weights and activations with the same precision.In this paper, we explore a new dimension of the design space: quantizing different layers with different bit-widths.We formulate this problem as a neural architecture search problem and propose a novel differentiable neural architecture search framework to efficiently explore its exponential search space with gradient-based optimization.Experiments show we surpass the state-of-the-art compression of ResNet on CIFAR-10 and ImageNet.Our quantized models with 21.1x smaller model size or 103.9x lower computational cost can still outperform baseline quantized or even full precision models.","A novel differentiable neural architecture search framework for mixed quantization of ConvNets.The authors introduce a new method for neural architecture search which selects the precision quantization of weights at each neural network layer, and use it in the context of network compression.The paper presents a new approach in network quantization by quantizing different layers with different bit-widths and introduces a new differentiable neural architecture search framework."
22,Smooth Loss Functions for Deep Top-k Classification,"The top- error is a common measure of performance in machine learning and computer vision.In practice, top- classification is typically performed with deep neural networks trained with the cross-entropy loss.Theoretical results indeed suggest that cross-entropy is an optimal learning objective for such a task in the limit of infinite data.In the context of limited and noisy data however, the use of a loss function that is specifically designed for top- classification can bring significant improvements.Our empirical evidence suggests that the loss function must be smooth and have non-sparse gradients in order to work well with deep neural networks.Consequently, we introduce a family of smoothed loss functions that are suited to top- optimization via deep learning.The widely used cross-entropy is a special case of our family.Evaluating our smooth loss functions is computationally challenging: a nave algorithm would require operations, where is the number of classes.Thanks to a connection to polynomial algebra and a divide-and-conquer approach, we provide an algorithm with a time complexity of.Furthermore, we present a novel approximation to obtain fast and stable algorithms on GPUs with single floating point precision.We compare the performance of the cross-entropy loss and our margin-based losses in various regimes of noise and data size, for the predominant use case of.Our investigation reveals that our loss is more robust to noise and overfitting than cross-entropy.","Smooth Loss Function for Top-k Error MinimizationProposes using top-k loss with deep models to address the problem of class confusion with similar classes both present or absent of the training dataset.Smoothes the top-k losses.This paper introduces a smooth surrogate loss function for the top-k SVM, for the purpose of plugging the SVM to the deep neural networks."
23,Mol-CycleGAN - a generative model for molecular optimization,"Designing a molecule with desired properties is one of the biggest challenges in drug development, as it requires optimization of chemical compound structures with respect to many complex properties.To augment the compound design process we introduce Mol-CycleGAN -- a CycleGAN-based model that generates optimized compounds with a chemical scaffold of interest.Namely, given a molecule our model generates a structurally similar one with an optimized value of the considered property.We evaluate the performance of the model on selected optimization objectives related to structural properties and to a physicochemical property.In the task of optimization of penalized logP of drug-like molecules our model significantly outperforms previous results.","We introduce Mol-CycleGAN - a new generative model for optimization of molecules to augment drug design.The paper presents an approach for optimizing molecular properties based on the application of CycleGANs to variational autoencoders for molecules and employs a domain-specific VAE called Junction Tree VAE (JT-VAE).This paper uses a variational autoencoders to learn a translation function, from the set of molecules without the interested property to the set of molecules with the property. "
24,Model Distillation with Knowledge Transfer from Face Classification to Alignment and Verification,"Knowledge distillation is a potential solution for model compression.The idea is to make a small student network imitate the target of a large teacher network, then the student network can be competitive to the teacher one.Most previous studies focus on model distillation in the classification task, where they propose different architectures and initializations for the student network.However, only the classification task is not enough, and other related tasks such as regression and retrieval are barely considered.To solve the problem, in this paper, we take face recognition as a breaking point and propose model distillation with knowledge transfer from face classification to alignment and verification.By selecting appropriate initializations and targets in the knowledge transfer, the distillation can be easier in non-classification tasks.Experiments on the CelebA and CASIA-WebFace datasets demonstrate that the student network can be competitive to the teacher one in alignment and verification, and even surpasses the teacher network under specific compression rates.In addition, to achieve stronger knowledge transfer, we also use a common initialization trick to improve the distillation performance of classification.Evaluations on the CASIA-Webface and large-scale MS-Celeb-1M datasets show the effectiveness of this simple trick.",We take face recognition as a breaking point and propose model distillation with knowledge transfer from face classification to alignment and verificationThis paper proposes to transfer the classifier from the model for face classification to the task of alignment and verification.The manuscript presents experiments on distilling knowledge from a face classification model to student models for face alignment and verification.
25,Recurrent Neural Networks with Top-k Gains for Session-based Recommendations,"RNNs have been shown to be excellent models for sequential data and in particular for session-based user behavior.The use of RNNs provides impressive performance benefits over classical methods in session-based recommendations.In this work we introduce a novel ranking loss function tailored for RNNs in recommendation settings.The better performance of such loss over alternatives, along with further tricks and improvements described in this work, allow to achieve an overall improvement of up to 35% in terms of MRR and Recall@20 over previous session-based RNN solutions and up to 51% over classical collaborative filtering approaches.Unlike data augmentation-based improvements, our method does not increase training times significantly.","Improving session-based recommendations with RNNs (GRU4Rec) by 35% using newly designed loss functions and sampling.This paper analyzes existing loss functions for session-based recommendations and proposes two novel losses functions which add a weighting to existing ranking-based loss functionsPresents modifications on top of earlier work for session-based recommendation using RNN by weighting negative examples by their ""relevance""This paper discusses the issues for optimizing the loss functions in GRU4Rec, proposes tricks for optimizing, and suggests an enhanced version."
26,Lifelong Learning by Adjusting Priors,"In representational lifelong learning an agent aims to continually learn to solve novel tasks while updating its representation in light of previous tasks.Under the assumption that future tasks are related to previous tasks, representations should be learned in such a way that they capture the common structure across learned tasks, while allowing the learner sufficient flexibility to adapt to novel aspects of a new task.We develop a framework for lifelong learning in deep neural networks that is based on generalization bounds, developed within the PAC-Bayes framework.Learning takes place through the construction of a distribution over networks based on the tasks seen so far, and its utilization for learning a new task.Thus, prior knowledge is incorporated through setting a history-dependent prior for novel tasks.We develop a gradient-based algorithm implementing these ideas, based on minimizing an objective function motivated by generalization bounds, and demonstrate its effectiveness through numerical examples.","We develop a lifelong learning approach to transfer learning based on PAC-Bayes theory, whereby priors are adjusted as new tasks are encountered thereby facilitating the learning of novel tasks.A novel PAC-Bayesian risk bound that serves as an objective function for multi-task machine learning, and an algorithm for minimizing a simplified version of that objective function.Extends existing PAC-Bayes bounds to multi-task learning, to allow the prior to be adapted across different tasks."
27,Normalized Direction-preserving Adam,"Optimization algorithms for training deep models not only affects the convergence rate and stability of the training process, but are also highly related to the generalization performance of trained models.While adaptive algorithms, such as Adam and RMSprop, have shown better optimization performance than stochastic gradient descent in many scenarios, they often lead to worse generalization performance than SGD, when used for training deep neural networks.In this work, we identify two problems regarding the direction and step size for updating the weight vectors of hidden units, which may degrade the generalization performance of Adam.As a solution, we propose the normalized direction-preserving Adam algorithm, which controls the update direction and step size more precisely, and thus bridges the generalization gap between Adam and SGD.Following a similar rationale, we further improve the generalization performance in classification tasks by regularizing the softmax logits.By bridging the gap between SGD and Adam, we also shed some light on why certain optimization algorithms generalize better than others.","A tailored version of Adam for training DNNs, which bridges the generalization gap between Adam and SGD.Proposes a variant of ADAM optimization algorithm that normalizes weights of each hidden unit using batch normalizationExtension of the Adam optimization algorithm to preserve the update direction by adapting the learning rate for the incoming weights to a hidden unit jointly using the L2 norm of the gradient vector"
28,Eigenoption Discovery through the Deep Successor Representation,"Options in reinforcement learning allow agents to hierarchically decompose a task into subtasks, having the potential to speed up learning and planning.However, autonomously learning effective sets of options is still a major challenge in the field.In this paper we focus on the recently introduced idea of using representation learning methods to guide the option discovery process.Specifically, we look at eigenoptions, options obtained from representations that encode diffusive information flow in the environment.We extend the existing algorithms for eigenoption discovery to settings with stochastic transitions and in which handcrafted features are not available. We propose an algorithm that discovers eigenoptions while learning non-linear state representations from raw pixels.It exploits recent successes in the deep reinforcement learning literature and the equivalence between proto-value functions and the successor representation.We use traditional tabular domains to provide intuition about our approach and Atari 2600 games to demonstrate its potential.","We show how we can use the successor representation to discover eigenoptions in stochastic domains, from raw pixels. Eigenoptions are options learned to navigate the latent dimensions of a learned representation.Extends the idea of eigenoptions to domains with stochastic transitions and where state features are learned.Shows equivalence between proto value functions and successor representations and derives the idea of eigen options as a mechanism in option discoveryThe paper is a follow up on previous work by Machado et al. (2017) showing how proto-value functions can be used to define options called “eigenoptions”."
29,Empirical Bounds on Linear Regions of Deep Rectifier Networks,"One form of characterizing the expressiveness of a piecewise linear neural network is by the number of linear regions, or pieces, of the function modeled.We have observed substantial progress in this topic through lower and upper bounds on the maximum number of linear regions and a counting procedure.However, these bounds only account for the dimensions of the network and the exact counting may take a prohibitive amount of time, therefore making it infeasible to benchmark the expressiveness of networks.In this work, we approximate the number of linear regions of specific rectifier networks with an algorithm for probabilistic lower bounds of mixed-integer linear sets.In addition, we present a tighter upper bound that leverages network coefficients.We test both on trained networks.The algorithm for probabilistic lower bounds is several orders of magnitude faster than exact counting and the values reach similar orders of magnitude, hence making our approach a viable method to compare the expressiveness of such networks.The refined upper bound is particularly stronger on networks with narrow layers. ","We provide improved upper bounds for the number of linear regions used in network expressivity, and an highly efficient algorithm (w.r.t. exact counting) to obtain probabilistic lower bounds on the actual number of linear regions.Contributes to the study of the number of linear regions in RELU neural networks by using an approximate probabilistic counting algorithm and analysisBuilds off previous work studying the counting of linear regions in deep neural networks, and improves the upper bound previously proposed by changing the dimensionality constraintThe paper deals with expressiveness of a piecewise linear neural network, characterized by the number of linear regions of the function modeled, and leverages probabilistic algorithms to compute the bounds faster, and proves tighter bounds."
30,A Biologically Inspired Visual Working Memory for Deep Networks,"The ability to look multiple times through a series of pose-adjusted glimpses is fundamental to human vision.This critical faculty allows us to understand highly complex visual scenes.Short term memory plays an integral role in aggregating the information obtained from these glimpses and informing our interpretation of the scene.Computational models have attempted to address glimpsing and visual attention but have failed to incorporate the notion of memory.We introduce a novel, biologically inspired visual working memory architecture that we term the Hebb-Rosenblatt memory.We subsequently introduce a fully differentiable Short Term Attentive Working Memory model which uses transformational attention to learn a memory over each image it sees.The state of our Hebb-Rosenblatt memory is embedded in STAWM as the weights space of a layer.By projecting different queries through this layer we can obtain goal-oriented latent representations for tasks including classification and visual reconstruction.Our model obtains highly competitive classification performance on MNIST and CIFAR-10.As demonstrated through the CelebA dataset, to perform reconstruction the model learns to make a sequence of updates to a canvas which constitute a parts-based representation.Classification with the self supervised representation obtained from MNIST is shown to be in line with the state of the art models.Finally, we show that STAWM can be trained under the dual constraints of classification and reconstruction to provide an interpretable visual sketchpad which helps open the `black-box' of deep learning.",A biologically inspired working memory that can be integrated in recurrent visual attention models for state of the art performanceIntroduces a new network architecture inspired by visual attentive working memory and applies it to classification tasks and using it as a generative modelThe paper augments the recurrent attention model with a novel Hebb-Rosenblatt working memory model and achieves competitive results on MNIST
31,Modulated Variational Auto-Encoders for Many-to-Many Musical Timbre Transfer,"Generative models have been successfully applied to image style transfer and domain translation.However, there is still a wide gap in the quality of results when learning such tasks on musical audio.Furthermore, most translation models only enable one-to-one or one-to-many transfer by relying on separate encoders or decoders and complex, computationally-heavy models.In this paper, we introduce the Modulated Variational auto-Encoders to perform musical timbre transfer.First, we define timbre transfer as applying parts of the auditory properties of a musical instrument onto another.We show that we can achieve and improve this task by conditioning existing domain translation techniques with Feature-wise Linear Modulation.Then, by replacing the usual adversarial translation criterion by a Maximum Mean Discrepancy objective, we alleviate the need for an auxiliary pair of discriminative networks.This allows a faster and more stable training, along with a controllable latent space encoder.By further conditioning our system on several different instruments, we can generalize to many-to-many transfer within a single variational architecture able to perform multi-domain transfers.Our models map inputs to 3-dimensional representations, successfully translating timbre from one instrument to another and supporting sound synthesis on a reduced set of control parameters.We evaluate our method in reconstruction and generation tasks while analyzing the auditory descriptor distributions across transferred domains.We show that this architecture incorporates generative controls in multi-domain transfer, yet remaining rather light, fast to train and effective on small datasets.","The paper uses Variational Auto-Encoding and network conditioning for Musical Timbre Transfer, we develop and generalize our architecture for many-to-many instrument transfers together with visualizations and evaluations.Proposes a Modulated Variational auto-Encoder to perform musical timbre transfer by replacing the usual adversarial translation criterion by a Maxiimum Mean DiscrepancyDescribes a many-to-many model for musical timbre transfer which builds on recent developments in domain and style transferProposes a hybrid VAE-based model to perform timbre transfer on recordings of musical instruments."
32,"On Random Deep Weight-Tied Autoencoders: Exact Asymptotic Analysis, Phase Transitions, and Implications to Training","We study the behavior of weight-tied multilayer vanilla autoencoders under the assumption of random weights.Via an exact characterization in the limit of large dimensions, our analysis reveals interesting phase transition phenomena when the depth becomes large.This, in particular, provides quantitative answers and insights to three questions that were yet fully understood in the literature.Firstly, we provide a precise answer on how the random deep weight-tied autoencoder model performs “approximate inference” as posed by Scellier et al., and its connection to reversibility considered by several theoretical studies.Secondly, we show that deep autoencoders display a higher degree of sensitivity to perturbations in the parameters, distinct from the shallow counterparts.Thirdly, we obtain insights on pitfalls in training initialization practice, and demonstrate experimentally that it is possible to train a deep autoencoder, even with the tanh activation and a depth as large as 200 layers, without resorting to techniques such as layer-wise pre-training or batch normalization.Our analysis is not specific to any depths or any Lipschitz activations, and our analytical techniques may have broader applicability.","We study the behavior of weight-tied multilayer vanilla autoencoders under the assumption of random weights. Via an exact characterization in the limit of large dimensions, our analysis reveals interesting phase transition phenomena.A theoretical analysis of autoencoders with weights tied between encoder and decoder (weight-tied) via mean field analysisAnalyses the performances of weighted tied auto-encoders by building on recent progress in analysis of high-dimensional statistics problems and specifically, the message passing algorithmThis paper studies auto-encoders under several assumptions, and points out that this model of random autoencoder can be elegantly and rigorously analysed with one-dimensional equations."
33,Discriminator Rejection Sampling,"We propose a rejection sampling scheme using the discriminator of a GAN toapproximately correct errors in the GAN generator distribution.We show thatunder quite strict assumptions, this will allow us to recover the data distributionexactly.We then examine where those strict assumptions break down and design apractical algorithm—called Discriminator Rejection Sampling—that can beused on real data-sets.Finally, we demonstrate the efficacy of DRS on a mixture ofGaussians and on the state of the art SAGAN model.On ImageNet, we train animproved baseline that increases the best published Inception Score from 52.52 to62.36 and reduces the Frechet Inception Distance from 18.65 to 14.79.We then useDRS to further improve on this baseline, improving the Inception Score to 76.08and the FID to 13.75.","We use a GAN discriminator to perform an approximate rejection sampling scheme on the output of the GAN generator. Proposes a rejection sampling algorithm for sampling from the GAN generator.This paper proposed a post-processing rejection sampling scheme for GANs, named Discriminator Rejection Sampling, to help filter ‘good’ samples from GANs’ generator."
34,Simple Fast Convolutional Feature Learning,"The quality of the features used in visual recognition is of fundamental importance for the overall system.For a long time, low-level hand-designed feature algorithms as SIFT and HOG have obtained the best results on image recognition.Visual features have recently been extracted from trained convolutional neural networks.Despite the high-quality results, one of the main drawbacks of this approach, when compared with hand-designed features, is the training time required during the learning process.In this paper, we propose a simple and fast way to train supervised convolutional models to feature extraction while still maintaining its high-quality.This methodology is evaluated on different datasets and compared with state-of-the-art approaches.",A simple fast method for extracting visual features from convolutional neural networksProposes a fast way to learn convolutional features that later can be used with any classifier by using reduced numbers of training epocs and specific schedule delays of learning rateUse a learning rate decay scheme that is fixed relative to the number of epochs used in training and extract the penultimate layer output as features to train a conventional classifier.
35,A Max-Affine Spline Perspective of Recurrent Neural Networks,"We develop a framework for understanding and improving recurrent neural networks using max-affine spline operators.We prove that RNNs using piecewise affine and convex nonlinearities can be written as a simple piecewise affine spline operator.The resulting representation provides several new perspectives for analyzing RNNs, three of which we study in this paper.First, we show that an RNN internally partitions the input space during training and that it builds up the partition through time.Second, we show that the affine slope parameter of an RNN corresponds to an input-specific template, from which we can interpret an RNN as performing a simple template matching given the input.Third, by carefully examining the MASO RNN affine mapping, we prove that using a random initial hidden state corresponds to an explicit L2 regularization of the affine parameters, which can mollify exploding gradients and improve generalization.Extensive experiments on several datasets of various modalities demonstrate and validate each of the above conclusions.In particular, using a random initial hidden states elevates simple RNNs to near state-of-the-art performers on these datasets.","We provide new insights and interpretations of RNNs from a max-affine spline operators perspective.Rewrites equations of Elman RNN in terms of so-called max-affine spline operatorsProvide a novel approach toward understanding RNNs using max-affline spline operators (MASO) by rewriting them with piecewise affine and convex activations MASOsThe authors build upon max-affine spline operator interpetation of a substantial class of deep networks, focusing on Recurrent Neural Networks using noise in initial hidden state acts as regularization"
36,Scalable Neural Theorem Proving on Knowledge Bases and Natural Language,"Reasoning over text and Knowledge Bases is a major challenge for Artificial Intelligence, with applications in machine reading, dialogue, and question answering. Transducing text to logical forms which can be operated on is a brittle and error-prone process. Operating directly on text by jointly learning representations and transformations thereof by means of neural architectures that lack the ability to learn and exploit general rules can be very data-inefficient and not generalise correctly. These issues are addressed by Neural Theorem Provers, neuro-symbolic systems based on a continuous relaxation of Prolog’s backward chaining algorithm, where symbolic unification between atoms is replaced by a differentiable operator computing the similarity between their embedding representations. In this paper, we first propose Neighbourhood-approximated Neural Theorem Provers consisting of two extensions toNTPs, namelya) a method for drastically reducing the previously prohibitive time and space complexity during inference and learning, andb) an attention mechanism for improving the rule learning process, deeming them usable on real-world datasets.Then, we propose a novel approach for jointly reasoning over KB facts and textual mentions, by jointly embedding them in a shared embedding space.The proposed method is able to extract rules and provide explanations—involving both textual patterns and KB relations—from large KBs and text corpora.We show that NaNTPs perform on par with NTPs at a fraction of a cost, and can achieve competitive link prediction results on challenging large-scale datasets, including WN18, WN18RR, and FB15k-237 while being able to provide explanations for each prediction and extract interpretable rules.","We scale Neural Theorem Provers to large datasets, improve the rule learning process, and extend it to jointly reason over text and Knowledge Bases.Proposes an extension of the Neural Theorem Provers system that addresses the main issues of this model by reducing the time and space complexity of the modelScales NTPs by using approximate nearest neighbour search over facts and rules during unification and suggests parameterizing predicates using attention over known predicatesimproves upon the previously proposed Neural Theorem Prover approach by using nearest neighbor search."
37,Generalization of Learning using Reservoir Computing,"We investigate the methods by which a Reservoir Computing Network learns concepts such as 'similar' and 'different' between pairs of images using a small training dataset and generalizes these concepts to previously unseen types of data.Specifically, we show that an RCN trained to identify relationships between image-pairs drawn from a subset of digits from the MNIST database or the depth maps of subset of visual scenes from a moving camera generalizes the learned transformations to images of digits unseen during training or depth maps of different visual scenes.We infer, using Principal Component Analysis, that the high dimensional reservoir states generated from an input image pair with a specific transformation converge over time to a unique relationship.Thus, as opposed to training the entire high dimensional reservoir state, the reservoir only needs to train on these unique relationships, allowing the reservoir to perform well with very few training examples.Thus, generalization of learning to unseen images is interpretable in terms of clustering of the reservoir state onto the attractor corresponding to the transformation in reservoir space.We find that RCNs can identify and generalize linear and non-linear transformations, and combinations of transformations, naturally and be a robust and effective image classifier.Additionally, RCNs perform significantly better than state of the art neural network classification techniques such as deep Siamese Neural Networks in generalization tasks both on the MNIST dataset and more complex depth maps of visual scenes from a moving camera.This work helps bridge the gap between explainable machine learning and biological learning through analogies using small datasets, and points to new directions in the investigation of learning processes.","Generalization of the relationships learnt between pairs of images using a small training data to previously unseen types of images using an explainable dynamical systems model, Reservoir Computing, and a biologically plausible learning technique based on analogies.Claims results of ""combining transformations"" in the context of RC by using an echo-state network with standard tanh acctivations with the difference that recurrent weights are not trainedNovel method of classifying different distorions of MNIST dataThe paper uses an echo state network to learn to classify image transformations between pairs of images into one of fives classes."
38,Generative Adversarial Models for Learning Private and Fair Representations,"We present Generative Adversarial Privacy and Fairness, a data-driven framework for learning private and fair representations of the data.GAPF leverages recent advances in adversarial learning to allow a data holder to learn ""universal"" representations that decouple a set of sensitive attributes from the rest of the dataset.Under GAPF, finding the optimal decorrelation scheme is formulated as a constrained minimax game between a generative decorrelator and an adversary.We show that for appropriately chosen adversarial loss functions, GAPF provides privacy guarantees against strong information-theoretic adversaries and enforces demographic parity.We also evaluate the performance of GAPF on multi-dimensional Gaussian mixture models and real datasets, and show how a designer can certify that representations learned under an adversary with a fixed architecture perform well against more complex adversaries.","We present Generative Adversarial Privacy and Fairness (GAPF), a data-driven framework for learning private and fair representations with certified privacy/fairness guaranteesThis paper uses a GAN model to provide an overview of the related work to Private/Fair Representation Learning (PRL).This paper presents an adversarial-based approach for private and fair representations by learned distortion of data that minimises the dependency on sensitive variables while the degree of distortion is constrained.The authors describe a framework of how to learn a demographic parity representation that can be used to train certain classifiers."
39,Evaluation Methodology for Attacks Against Confidence Thresholding Models,"Current machine learning algorithms can be easily fooled by adversarial examples.One possible solution path is to make models that use confidence thresholding to avoid making mistakes.Such models refuse to make a prediction when they are not confident of their answer.We propose to evaluate such models in terms of tradeoff curves with the goal of high success rate on clean examples and low failure rate on adversarial examples.Existing untargeted attacks developed for models that do not use confidence thresholding tend to underestimate such models' vulnerability.We propose the MaxConfidence family of attacks, which are optimal in a variety of theoretical settings, including one realistic setting: attacks against linear models.Experiments show the attack attains good results in practice.We show that simple defenses are able to perform well on MNIST but not on CIFAR, contributing further to previous calls that MNIST should be retired as a benchmarking dataset for adversarial robustness research. We release code for these evaluations as part of the cleverhans library .","We present metrics and an optimal attack for evaluating models that defend against adversarial examples using confidence thresholdingThis paper introduces a family of attack on confidence thresholding algortihms, focusing mainly on evaluation methodologies.Proposes an evaluation method for confidence thresholding defense models and an approach for generating adversarial examples by choosing the wrong class with the most confidence when using targeted attacksThe paper presents an evaluation methodology for evaluating attacks on confidence thresholding methods and proposes a new kind of attack."
40,Competitive experience replay,"Deep learning has achieved remarkable successes in solving challenging reinforcement learning problems when dense reward function is provided.However, in sparse reward environment it still often suffers from the need to carefully shape reward function to guide policy optimization.This limits the applicability of RL in the real world since both reinforcement learning and domain-specific knowledge are required.It is therefore of great practical importance to develop algorithms which can learn from a binary signal indicating successful task completion or other unshaped, sparse reward signals.We propose a novel method called competitive experience replay, which efficiently supplements a sparse reward by placing learning in the context of an exploration competition between a pair of agents.Our method complements the recently proposed hindsight experience replay by inducing an automatic exploratory curriculum.We evaluate our approach on the tasks of reaching various goal locations in an ant maze and manipulating objects with a robotic arm.Each task provides only binary rewards indicating whether or not the goal is achieved.Our method asymmetrically augments these sparse rewards for a pair of agents each learning the same task, creating a competitive game designed to drive exploration.Extensive experiments demonstrate that this method leads to faster converge and improved task performance.","a novel method to learn with sparse reward using adversarial reward re-labelingProposes to use a competitive multi-agent setting for encouraging exploration and shows that CER + HER > HER ~ CERPropose a new method for learning from sparse rewards in model-free reinforcement learning settings and densifying rewardTo address the sparse reward problems and encourage exploration in RL algorithms, the authors propose a relabeling strategy called Competitive Experience Reply (CER)."
41,Hierarchical Generative Modeling for Controllable Speech Synthesis,"This paper proposes a neural end-to-end text-to-speech model which can control latent attributes in the generated speech that are rarely annotated in the training data, such as speaking style, accent, background noise, and recording conditions.The model is formulated as a conditional generative model with two levels of hierarchical latent variables.The first level is a categorical variable, which represents attribute groups and provides interpretability.The second level, conditioned on the first, is a multivariate Gaussian variable, which characterizes specific attribute configurations and enables disentangled fine-grained control over these attributes.This amounts to using a Gaussian mixture model for the latent distribution.Extensive evaluation demonstrates its ability to control the aforementioned attributes.In particular, it is capable of consistently synthesizing high-quality clean speech regardless of the quality of the training data for the target speaker.","Building a TTS model with Gaussian Mixture VAEs enables fine-grained control of speaking style, noise condition, and more.Describes the conditioned GAN model to generate speaker conditioned Mel spectra by augmenting the z-space corresponding to the identificationThis paper proposes a two layer latent variable model to obtain disentangled latent representation, thus facilitating fine-grained control over various attributesThis paper proposes a model that can control non-annotated attributes such as speaking style, accent, background noise, etc."
42,Learning to Count Objects in Natural Images for Visual Question Answering,"Visual Question Answering models have struggled with counting objects in natural images so far.We identify a fundamental problem due to soft attention in these models as a cause.To circumvent this problem, we propose a neural network component that allows robust counting from object proposals.Experiments on a toy task show the effectiveness of this component and we obtain state-of-the-art accuracy on the number category of the VQA v2 dataset without negatively affecting other categories, even outperforming ensemble models with our single model.On a difficult balanced pair metric, the component gives a substantial improvement in counting over a strong baseline by 6.6%.","Enabling Visual Question Answering models to count by handling overlapping object proposals.This paper proposes a hand-designed network architecture on a graph of object proposals to perform soft non-maximum suppression to get object count.Focuses on a counting problem in visual question answering using attention mechanism and propsoe a differentiable counting compoent which explicitly counts the number of objectsThis paper tackles the object counting problem in visual question answering, it proposes many heuristics to find the correct count."
43,Zero-training Sentence Embedding via Orthogonal Basis,"We propose a simple and robust training-free approach for building sentence representations.Inspired by the Gram-Schmidt Process in geometric theory, we build an orthogonal basis of the subspace spanned by a word and its surrounding context in a sentence.We model the semantic meaning of a word in a sentence based on two aspects.One is its relatedness to the word vector subspace already spanned by its contextual words.The other is its novel semantic meaning which shall be introduced as a new basis vector perpendicular to this existing subspace. Following this motivation, we develop an innovative method based on orthogonal basis to combine pre-trained word embeddings into sentence representation.This approach requires zero training and zero parameters, along with efficient inference performance.We evaluate our approach on 11 downstream NLP tasks.Experimental results show that our model outperforms all existing zero-training alternatives in all the tasks and it is competitive to other approaches relying on either large amounts of labelled data or prolonged training time.","A simple and training-free approach for sentence embeddings with competitive performance compared with sophisticated models requiring either large amount of training data or prolonged training time.Presented a new training-free way of generating sentence embedding with systematic analysisProposes a new geometry-based method for sentence embedding from word embedding vectors by quantifying the novelty, significance, and corpus-uniqueness of each wordThis paper explores sentence embedding based on orthogonal decomposition of the spanned space by word embeddings"
44,Meta-Learning for Semi-Supervised Few-Shot Classification,"In few-shot classification, we are interested in learning algorithms that train a classifier from only a handful of labeled examples.Recent progress in few-shot classification has featured meta-learning, in which a parameterized model for a learning algorithm is defined and trained on episodes representing different classification problems, each with a small labeled training set and its corresponding test set.In this work, we advance this few-shot classification paradigm towards a scenario where unlabeled examples are also available within each episode.We consider two situations: one where all unlabeled examples are assumed to belong to the same set of classes as the labeled examples of the episode, as well as the more challenging situation where examples from other distractor classes are also provided.To address this paradigm, we propose novel extensions of Prototypical Networks that are augmented with the ability to use unlabeled examples when producing prototypes.These models are trained in an end-to-end way on episodes, to learn to leverage the unlabeled examples successfully.We evaluate these methods on versions of the Omniglot and miniImageNet benchmarks, adapted to this new framework augmented with unlabeled examples.We also propose a new split of ImageNet, consisting of a large set of classes, with a hierarchical structure.Our experiments confirm that our Prototypical Networks can learn to improve their predictions due to unlabeled examples, much like a semi-supervised algorithm would.","We propose novel extensions of Prototypical Networks that are augmented with the ability to use unlabeled examples when producing prototypes.This paper is an extension of a prototypical network that considers employing the unlabeled examples available to help train each episodeStudies the problem of semi-supervised few-shot classification by extending the prototypical networks into the setting of semi-supervised learning with example from distractor classesExtends the Prototypical Network to the semi-supervised setting by updating prototypes using assigned pseudo-labels, dealing with distractors, and weighing samples using distance to the original prototypes."
45,Distribution-Interpolation Trade off in Generative Models,"We investigate the properties of multidimensional probability distributions in the context of latent space prior distributions of implicit generative models.Our work revolves around the phenomena arising while decoding linear interpolations between two random latent vectors -- regions of latent space in close proximity to the origin of the space are oversampled, which restricts the usability of linear interpolations as a tool to analyse the latent space.We show that the distribution mismatch can be eliminated completely by a proper choice of the latent probability distribution or using non-linear interpolations.We prove that there is a trade off between the interpolation being linear, and the latent distribution having even the most basic properties required for stable training, such as finite mean.We use the multidimensional Cauchy distribution as an example of the prior distribution, and also provide a general method of creating non-linear interpolations, that is easily applicable to a large family of commonly used latent distributions.","We theoretically prove that linear interpolations are unsuitable for analysis of trained implicit generative models. Studies the problem of when the linear interpolant between two random variables follows the same distribution, related to prior distribution of an implicit generative modelThis work asks how to interpolate in the latent space given a latent variable model."
46,End-to-End Abnormality Detection in Medical Imaging,"Deep neural networks have shown promising performance in computer vision.In medical imaging, encouraging results have been achieved with deep learning for applications such as segmentation, lesion detection and classification.Nearly all of the deep learning based image analysis methods work on reconstructed images, which are obtained from original acquisitions via solving inverse problems.The reconstruction algorithms are designed for human observers, but not necessarily optimized for DNNs which can often observe features that are incomprehensible for human eyes.Hence, it is desirable to train the DNNs directly from the original data which lie in a different domain with the images.In this paper, we proposed an end-to-end DNN for abnormality detection in medical imaging.To align the acquisition with the annotations made by radiologists in the image domain, a DNN was built as the unrolled version of iterative reconstruction algorithms to map the acquisitions to images, and followed by a 3D convolutional neural network to detect the abnormality in the reconstructed images.The two networks were trained jointly in order to optimize the entire DNN for the detection task from the original acquisitions.The DNN was implemented for lung nodule detection in low-dose chest computed tomography, where a numerical simulation was done to generate acquisitions from 1,018 chest CT images with radiologists' annotations.The proposed end-to-end DNN demonstrated better sensitivity and accuracy for the task compared to a two-step approach, in which the reconstruction and detection DNNs were trained separately.A significant reduction of false positive rate on suspicious lesions were observed, which is crucial for the known over-diagnosis in low-dose lung CT imaging.The images reconstructed by the proposed end-to-end network also presented enhanced details in the region of interest.",Detection of lung nodule starting from projection data rather than images.DNNs are used for patch-based lung nodule detection in CT projection data.Jointly modeling computed tomography reconstruction and lesion detection in the lung by training the mapping from raw sinogram to detection outputs in an end-to-end mannerPresents an end to end training of a CNN architecture that combines CT image signal processing and image analysis.
47,Do Deep Reinforcement Learning Algorithms really Learn to Navigate?,"Deep reinforcement learning algorithms have demonstrated progress in learning to find a goal in challenging environments.As the title of the paper by Mirowski et al. suggests, one might assume that DRL-based algorithms are able to “learn to navigate” and are thus ready to replace classical mapping and path-planning algorithms, at least in simulated environments.Yet, from experiments and analysis in this earlier work, it is not clear what strategies are used by these algorithms in navigating the mazes and finding the goal.In this paper, we pose and study this underlying question: are DRL algorithms doing some form of mapping and/or path-planning?Our experiments show that the algorithms are not memorizing the maps of mazes at the testing stage but, rather, at the training stage.Hence, the DRL algorithms fall short of qualifying as mapping or path-planning algorithms with any reasonable definition of mapping.We extend the experiments in Mirowski et al. by separating the set of training and testing maps and by a more ablative coverage of the space of experiments.Our systematic experiments show that the NavA3C-D1-D2-L algorithm, when trained and tested on the same maps, is able to choose the shorter paths to the goal.However, when tested on unseen maps the algorithm utilizes a wall-following strategy to find the goal without doing any mapping or path planning.",We quantitatively and qualitatively evaluate deep reinforcement learning based navigation methods under a variety of conditions to answer the question of how close they are to replacing classical path planners and mapping algorithms.Evaluate a Deep RL-based model on training mazes by measuring repeated latency to goal and comparison to shortest route
48,On Learning Heteroscedastic Noise Models within Differentiable Bayes Filters,"In many robotic applications, it is crucial to maintain a belief about the state ofa system, like the location of a robot or the pose of an object.These state estimates serve as input for planning and decision making andprovide feedback during task execution.Recursive Bayesian Filtering algorithms address the state estimation problem,but they require a model of the process dynamics and the sensory observations as well asnoise estimates that quantify the accuracy of these models.Recently, multiple works have demonstrated that the process and sensor models can belearned by end-to-end training through differentiable versions of Recursive Filtering methods.However, even if the predictive models are known, finding suitable noise modelsremains challenging.Therefore, many practical applications rely on very simplistic noisemodels.Our hypothesis is that end-to-end training through differentiable BayesianFilters enables us to learn more complex heteroscedastic noise models forthe system dynamics.We evaluate learning such models with different types offiltering algorithms and on two different robotic tasks.Our experiments show that especiallyfor sampling-based filters like the Particle Filter, learning heteroscedastic noisemodels can drastically improve the tracking performance in comparison to usingconstant noise models.",We evaluate learning heteroscedastic noise models within different Differentiable Bayes FiltersProposes to learn heteroscedastic noise models from data by optimizing the prediction likelihood end-toend through differentiable Bayesian Filters and two different versions of the Unscented Kalman FilterRevisits Bayes filters and evaluates the benefit of training the observation and process noise models while keeping all other models fixedThis paper presents a method to learn and use state and observation dependent noise in traditional Bayesian filtering algorithms. The approach consists of constructing a neural network model which takes as input the raw observation data and produces a compact representation and an associated diagonal covariance.
49,Rethinking Knowledge Graph Propagation for Zero-Shot Learning,"Graph convolutional neural networks have recently shown great potential for the task of zero-shot learning.These models are highly sample efficient as related concepts in the graph structure share statistical strength allowing generalization to new classes when faced with a lack of data.However, we find that the extensive use of Laplacian smoothing at each layer in current approaches can easily dilute the knowledge from distant nodes and consequently decrease the performance in zero-shot learning.In order to still enjoy the benefit brought by the graph structure while preventing the dilution of knowledge from distant nodes, we propose a Dense Graph Propagation module with carefully designed direct links among distant nodes.DGP allows us to exploit the hierarchical graph structure of the knowledge graph through additional connections.These connections are added based on a node's relationship to its ancestors and descendants.A weighting scheme is further used to weigh their contribution depending on the distance to the node.Combined with finetuning of the representations in a two-stage training approach our method outperforms state-of-the-art zero-shot learning approaches.","We rethink the way information can be exploited more efficiently in the knowledge graph in order to improve performance on the Zero-Shot Learning task and propose a dense graph propagation (DGP) module for this purpose.This authors propose a solution to the problem of over-smoothing in Graph conv networks by allowing dense propagation between all related nodes, weighted by the mutual distance.Proposes a novel graph convolutional neural network to tackle the problem of zero-shot classification by using relational structures between classes as input of graph convolutional networks to learn classifiers of unseen classes"
50,Trace-back along capsules and its application on semantic segmentation  		,"In this paper, we propose a capsule-based neural network model to solve the semantic segmentation problem.By taking advantage of the extractable part-whole dependencies available in capsule layers, we derive the probabilities of the class labels for individual capsules through a recursive, layer-by-layer procedure.We model this procedure as a traceback pipeline and take it as a central piece to build an end-to-end segmentation network.Under the proposed framework, image-level class labels and object boundaries are jointly sought in an explicit manner, which poses a significant advantage over the state-of-the-art fully convolutional network solutions.Experiments conducted on modified MNIST and neuroimages demonstrate that our model considerably enhance the segmentation performance compared to the leading FCN variant.","A capsule-based semantic segmentation, in which the probabilities of the class labels are traced back through capsule pipeline. The authors present a trace-back mechanism to associate lowest level of Capsules with their respective classesProposes a traceback layer for capsule networks to do semantic segmentation and makes explicit use of part-whole relationship in the capsule layersProposes a trace-back method based on the CapsNet concept of Sabour to perform a semantic segmentation in parallel to classification."
51,On the Trajectory of Stochastic Gradient Descent in the Information Plane,"Studying the evolution of information theoretic quantities during Stochastic Gradient Descent learning of Artificial Neural Networks has gained popularity in recent years.Nevertheless, these type of experiments require estimating mutual information and entropy which becomes intractable for moderately large problems.In this work we propose a framework for understanding SGD learning in the information plane which consists of observing entropy and conditional entropy of the output labels of ANN.Through experimental results and theoretical justifications it is shown that, under some assumptions, the SGD learning trajectories appear to be similar for different ANN architectures.First, the SGD learning is modeled as a Hidden Markov Process whose entropy tends to increase to the maximum.Then, it is shown that the SGD learning trajectory appears to move close to the shortest path between the initial and final joint distributions in the space of probability measures equipped with the total variation metric.Furthermore, it is shown that the trajectory of learning in the information plane can provide an alternative for observing the learning process, with potentially richer information about the learning than the trajectories in training and test error.","We look at SGD as a trajectory in the space of probability measures, show its connection to Markov processes, propose a simple Markov model of SGD learning, and experimentally compare it with SGD using information theoretic quantities. Constructs a Markov chain that follows a shorted path in TV metric on P and shows that trajectories of SGD and alpha-SMLC have similar conditional entropyStudies the trajectory of H(hat) versus H(hat|y) on the information plane for stochastic gradient descent methods for training neural networks Describes SGD from the point of view of the distribution p(y',y) where y is (a possibly corrupted) true class-label and y' a model prediction."
52,Meta-Learning For Stochastic Gradient MCMC,"Stochastic gradient Markov chain Monte Carlo has become increasingly popular for simulating posterior samples in large-scale Bayesian modeling.However, existing SG-MCMC schemes are not tailored to any specific probabilistic model, even a simple modification of the underlying dynamical system requires significant physical intuition.This paper presents the first meta-learning algorithm that allows automated design for the underlying continuous dynamics of an SG-MCMC sampler.The learned sampler generalizes Hamiltonian dynamics with state-dependent drift and diffusion, enabling fast traversal and efficient exploration of energy landscapes.Experiments validate the proposed approach on Bayesian fully connected neural network, Bayesian convolutional neural network and Bayesian recurrent neural network tasks, showing that the learned sampler outperforms generic, hand-designed SG-MCMC algorithms, and generalizes to different datasets and larger architectures.",This paper proposes a method to automate the design of stochastic gradient MCMC proposal using meta learning approach. Prsents a meta-learning approach to automatically design MCMC sampler based on Hamiltonian dynamics to mix faster on problems similar to training problemsParameterizes diffusion and curl matrices by neural networks and meta-learn and optimize an sg-mcmc algorithm. 
53,Image Quality Assessment Techniques Improve Training and Evaluation of Energy-Based Generative Adversarial Networks,"We propose a new, multi-component energy function for energy-based Generative Adversarial Networks based on methods from the image quality assessment literature.Our approach expands on the Boundary Equilibrium Generative Adversarial Network by outlining some of the short-comings of the original energy and loss functions.We address these short-comings by incorporating an l1 score, the Gradient Magnitude Similarity score, and a chrominance score into the new energy function.We then provide a set of systematic experiments that explore its hyper-parameters.We show that each of the energy function's components is able to represent a slightly different set of features, which require their own evaluation criteria to assess whether they have been adequately learned.We show that models using the new energy function are able to produce better image representations than the BEGAN model in predicted ways.","Image Quality Assessment Techniques Improve Training and Evaluation of Energy-Based Generative Adversarial NetworksProposes an energy-based formulation to the BEGAN modeal and modifies it to include an image quality assessment based termProposes some new energy function in the BEGAN (boundary equilibrium GAN framework), including l_1 score, Gradient magnitude similarity score, and chrominance score."
54,Aggregated Momentum: Stability Through Passive Damping,"Momentum is a simple and widely used trick which allows gradient-based optimizers to pick up speed along low curvature directions.Its performance depends crucially on a damping coefficient.Largecamping coefficients can potentially deliver much larger speedups, but are prone to oscillations and instability; hence one typically resorts to small values such as 0.5 or 0.9.We propose Aggregated Momentum, a variant of momentum which combines multiple velocity vectors with different damping coefficients.AggMo is trivial to implement, but significantly dampens oscillations, enabling it to remain stable even for aggressive damping coefficients such as 0.999.We reinterpret Nesterov's accelerated gradient descent as a special case of AggMo and analyze rates of convergence for quadratic objectives.Empirically, we find that AggMo is a suitable drop-in replacement for other momentum methods, and frequently delivers faster convergence with little to no tuning.","We introduce a simple variant of momentum optimization which is able to outperform classical momentum, Nesterov, and Adam on deep learning tasks with minimal hyperparameter tuning.Introduces a variant of momentum that aggregates several velocities with different dampening coefficients that significantly decreases oscillationProposed an aggregated momentum methods for gradient based optimization by using multiple velocity vectors with different damping factors instead of a single velocity vector to improve stabilityThe authors combine several update steps together to achieve aggregated momentum also demonstrating it is more stable than the other momentum methods"
55,Efficiently applying attention to sequential data with the Recurrent Discounted Attention unit,"Recurrent Neural Networks architectures excel at processing sequences bymodelling dependencies over different timescales.The recently introducedRecurrent Weighted Average unit captures long term dependenciesfar better than an LSTM on several challenging tasks.The RWA achievesthis by applying attention to each input and computing a weighted averageover the full history of its computations.Unfortunately, the RWA cannotchange the attention it has assigned to previous timesteps, and so struggleswith carrying out consecutive tasks or tasks with changing requirements.We present the Recurrent Discounted Attention unit that builds onthe RWA by additionally allowing the discounting of the past.We empirically compare our model to RWA, LSTM and GRU units onseveral challenging tasks.On tasks with a single output the RWA, RDA andGRU units learn much quicker than the LSTM and with better performance.On the multiple sequence copy task our RDA unit learns the task threetimes as quickly as the LSTM or GRU units while the RWA fails to learn atall.On the Wikipedia character prediction task the LSTM performs bestbut it followed closely by our RDA unit.Overall our RDA unit performswell and is sample efficient on a large variety of sequence tasks.","We introduce the Recurrent Discounted Unit which applies attention to any length sequence in linear timeThis paper proposes the Recurrent Discounted Attention (RDA), an extension to Recurrent Weighted Average (RWA) by adding a discount factor.Extends the recurrent weight average to overcome the limitation of the original method while maintaining its advantage and proposes the method of using Elman nets as the base RNN"
56,Variance Networks: When Expectation Does Not Meet Your Expectations,"Ordinary stochastic neural networks mostly rely on the expected values of their weights to make predictions, whereas the induced noise is mostly used to capture the uncertainty, prevent overfitting and slightly boost the performance through test-time averaging.In this paper, we introduce variance layers, a different kind of stochastic layers.Each weight of a variance layer follows a zero-mean distribution and is only parameterized by its variance.It means that each object is represented by a zero-mean distribution in the space of the activations.We show that such layers can learn surprisingly well, can serve as an efficient exploration tool in reinforcement learning tasks and provide a decent defense against adversarial attacks.We also show that a number of conventional Bayesian neural networks naturally converge to such zero-mean posteriors.We observe that in these cases such zero-mean parameterization leads to a much better training objective than more flexible conventional parameterizations where the mean is being learned.","It is possible to learn a zero-centered Gaussian distribution over the weights of a neural network by learning only variances, and it works surprisingly well.This paper investigates the effects of mean of variational posterior and proposes variance layer, which only uses variance to store informationStudies variance neural networks which approximate the posterior of Bayesian neural networks with zero-mean Gaussian distributions"
57,Network of Graph Convolutional Networks Trained on Random Walks,"Graph Convolutional Networks are a recently proposed architecture which has had success in semi-supervised learning on graph-structured data.At the same time, unsupervised learning of graph embeddings has benefited from the information contained in random walks.In this paper we propose a model, Network of GCNs, which marries these two lines of work.At its core, N-GCN trains multiple instances of GCNs over node pairs discovered at different distances in random walks, and learns a combination of the instance outputs which optimizes the classification objective.Our experiments show that our proposed N-GCN model achieves state-of-the-art performance on all of the challenging node classification tasks we consider: Cora, Citeseer, Pubmed, and PPI.In addition, our proposed method has other desirable properties, including generalization to recently proposed semi-supervised learning methods such as GraphSAGE, allowing us to propose N-SAGE, and resilience to adversarial input perturbations.","We make a network of Graph Convolution Networks, feeding each a different power of the adjacency matrix, combining all their representation into a classification sub-network, achieving state-of-the-art on semi-supervised node classification.Proposes a new network of GCNs with two approaches: a fully connected layer on top of stacked features and attention mechanism that uses scalar weight per GCN.Presents a Network of Graph Convolutional Networks that uses random walk statistics to extract information from near and distant neighbors in the graph"
58,Cheap DNN Pruning with Performance Guarantees ,Recent DNN pruning algorithms have succeeded in reducing the number of parameters in fully connected layers often with little or no drop in classification accuracy.However most of the existing pruning schemes either have to be applied during training or require a costly retraining procedure after pruning to regain classification accuracy.In this paper we propose a cheap pruning algorithm based on difference of convex optimisation.We also provide theoretical analysis for the growth in the Generalisation Error of the new pruned network.Our method can be used with any convex regulariser and allows for a controlled degradation in classification accuracy while being orders of magnitude faster than competing approaches.Experiments on common feedforward neural networks show that for sparsity levels above 90% our method achieves 10% higher classification accuracy compared to Hard Thresholding.,A fast pruning algorithm for fully connected DNN layers with theoretical analysis of degradation in Generalisation Error.Presents a cheap pruning algorithm for dense layers of DNNs.Proposes a solution to the problem of pruning DNNs by posing the Net-trim objective function as a Difference of convex(DC) function.
59,Video Action Segmentation with Hybrid Temporal Networks,"Action segmentation as a milestone towards building automatic systems to understand untrimmed videos has received considerable attention in the recent years.It is typically being modeled as a sequence labeling problem but contains intrinsic and sufficient differences than text parsing or speech processing.In this paper, we introduce a novel hybrid temporal convolutional and recurrent network, which has an encoder-decoder architecture: the encoder consists of a hierarchy of temporal convolutional kernels that capture the local motion changes of different actions; the decoder is a hierarchy of recurrent neural networks that are able to learn and memorize long-term action dependencies after the encoding stage.Our model is simple but extremely effective in terms of video sequence labeling.The experimental results on three public action segmentation datasets have shown that the proposed model achieves superior performance over the state of the art.","We propose a new hybrid temporal network that achieves state-of-the-art performance on video action segmentation on three public datasets.Discusses the problem of action segmentation in long videos, up to 10 minutes long by using a temporal convolutional encoder-decoder architectureProposes a combination of temporal convolutional and recurrent network for video action segmentation"
60,Feature Matters: A Stage-by-Stage Approach for Task Independent Knowledge Transfer,"Convolutional Neural Networks become deeper and deeper in recent years, making the study of model acceleration imperative.It is a common practice to employ a shallow network, called student, to learn from a deep one, which is termed as teacher.Prior work made many attempts to transfer different types of knowledge from teacher to student, however, there are two problems remaining unsolved.Firstly, the knowledge used by existing methods is highly dependent on task and dataset, limiting their applications.Secondly, there lacks an effective training scheme for the transfer process, leading to degradation of performance.In this work, we argue that feature is the most important knowledge from teacher.It is sufficient for student to just learn good features regardless of the target task.From this discovery, we further present an efficient learning strategy to mimic features stage by stage.Extensive experiments demonstrate the importance of features and show that the proposed approach significantly narrows down the gap between student and teacher, outperforming the state-of-the-art methods.",This paper proposes to transfer knowledge from deep model to shallow one by mimicking features stage by stage.Explains a stage by stage knowledge transer method by using different structures of resnetsThis paper proposes dividing a network into multiple parts and distilling each part sequentially to improve distillation performance in deep teacher networks
61,Improved robustness to adversarial examples using Lipschitz regularization of the loss,"We augment adversarial training with worst case adversarial training which improves adversarial robustness by 11% over the current state-of-the-art result in the `2-norm on CIFAR-10.We interpret adversarial training asTotal Variation Regularization, which is a fundamental tool in mathematical im-age processing, and WCAT as Lipschitz regularization, which appears in ImageInpainting.We obtain verifiable worst and average case robustness guarantees,based on the expected and maximum values of the norm of the gradient of theloss.","Improvements to adversarial robustness, as well as provable robustness guarantees, are obtained by augmenting adversarial training with a tractable Lipschitz regularizationExplores augmenting the training loss with an additional gradient regularization term to improve robustness of models against adversarial examplesUses a trick to simplify the adversarial loss by one in which the adversarial perturbation appears in closed form."
62,ElimiNet: A Model for Eliminating Options for Reading Comprehension with Multiple Choice Questions,"The task of Reading Comprehension with Multiple Choice Questions, requires a human to read a given } pair and select one of the given options.The current state of the art model for this task first computes a query-aware representation for the passage and then the option which has the maximum similarity with this representation.However, when humans perform this task they do not just focus on option selection but use a combination of and . Specifically, a human would first try to eliminate the most irrelevant option and then read the document again in the light of this new information.This process could be repeated multiple times till the reader is finally ready to select the correct option.We propose , a neural network based model which tries to mimic this process.Specifically, it has gates which decide whether an option can be eliminated given the } pair and if so it tries to make the document representation orthogonal to this eliminatedd option.The model makes multiple rounds of partial elimination to refine the document representation and finally uses a selection module to pick the best option.We evaluate our model on the recently released large scale RACE dataset and show that it outperforms the current state of the art model on 7 out of the 13 question types in this dataset.Further we show that taking an ensemble of our based method with a based method gives us an improvement of 7% over the best reported performance on this dataset. ",A model combining elimination and selection for answering multiple choice questionsGives an elaboration on the Gated Attention Reader adding gates based on answer elimination in multiple choice reading comprehensionThis paper proposes the use of an elimination gate in model architectures for reading comprehension tasks but does not achieve state-of-the-art resultsThis paper propses a new multi-choice reading comprehension model based on the idea that some options should be eliminated to infer better passage/question representations.
63,Probabilistic Recursive Reasoning for Multi-Agent Reinforcement Learning,"Humans are capable of attributing latent mental contents such as beliefs, or intentions to others.The social skill is critical in everyday life to reason about the potential consequences of their behaviors so as to plan ahead.It is known that humans use this reasoning ability recursively, i.e. considering what others believe about their own beliefs. In this paper, we start from level- recursion and introduce a probabilistic recursive reasoning framework for multi-agent reinforcement learning.Our hypothesis is that it is beneficial for each agent to account for how the opponents would react to its future behaviors.Under the PR2 framework, we adopt variational Bayes methods to approximate the opponents' conditional policy, to which each agent finds the best response and then improve their own policy.We develop decentralized-training-decentralized-execution algorithms, PR2-Q and PR2-Actor-Critic, that are proved to converge in the self-play scenario when there is one Nash equilibrium.Our methods are tested on both the matrix game and the differential game, which have a non-trivial equilibrium where common gradient-based methods fail to converge.Our experiments show that it is critical to reason about how the opponents believe about what the agent believes.We expect our work to contribute a new idea of modeling the opponents to the multi-agent reinforcement learning community. ","We proposed a novel probabilisitic recursive reasoning (PR2) framework for multi-agent deep reinforcement learning tasks.Proposes a new approach for fully decentralized training in multi-agent reinforcement learningTackles the problem of endowing RL agents with recursive reasoning capabilities in a multi-agent setting based on the hypothesis that recursive reasoning is beneficial for them to converge to non-trival equilibriaThe paper introduces a decentralized training method for multi-agent reinforcement learning, where the agents infer the policies of other agents and use the inferred models for decision making. "
64,Variance-based Gradient Compression for Efficient Distributed Deep Learning,"Due to the substantial computational cost, training state-of-the-art deep neural networks for large-scale datasets often requires distributed training using multiple computation workers.However, by nature, workers need to frequently communicate gradients, causing severe bottlenecks, especially on lower bandwidth connections.A few methods have been proposed to compress gradient for efficient communication, but they either suffer a low compression ratio or significantly harm the resulting model accuracy, particularly when applied to convolutional neural networks.To address these issues, we propose a method to reduce the communication overhead of distributed deep learning.Our key observation is that gradient updates can be delayed until an unambiguous gradient has been calculated.We also present an efficient algorithm to compute the variance and prove that it can be obtained with negligible additional cost.We experimentally show that our method can achieve very high compression ratio while maintaining the result model accuracy.We also analyze the efficiency using computation and communication cost models and provide the evidence that this method enables distributed deep learning for many scenarios with commodity environments.",A new algorithm to reduce the communication overhead of distributed deep learning by distinguishing ‘unambiguous’ gradients.Proposes a variance-based gradient compression method to reducee the communication overhead of distributed deep learningProposes a novel way of compressing gradient updates for distributed SGD in order to speed up overall executionIntroduces variance-based gradient compression method for efficient distributed training of neural networks and measuring ambuiguity.
65,Minimal-Entropy Correlation Alignment for Unsupervised Deep Domain Adaptation,"In this work, we face the problem of unsupervised domain adaptation with a novel deep learning approach which leverages our finding that entropy minimization is induced by the optimal alignment of second order statistics between source and target domains.We formally demonstrate this hypothesis and, aiming at achieving an optimal alignment in practical cases, we adopt a more principled strategy which, differently from the current Euclidean approaches, deploys alignment along geodesics.Our pipeline can be implemented by adding to the standard classification loss, a source-to-target regularizer that is weighted in an unsupervised and data-driven fashion.We provide extensive experiments to assess the superiority of our framework on standard domain and modality adaptation benchmarks.","A new unsupervised deep domain adaptation technique which efficiently unifies correlation alignment and entropy minimizationImproves the correlation alignment approach to domain adaptation by replacing the Euclidean distance with the geodesic Log-Euclidean distance between two covariance matices, and automatically selecting the balancing cost by the entropy on the target domain.Proposal for minimal-entropy correlation alignment, an unsupervised domain adaptation algorithm which links together entropy minimization and correlation alignment methods."
66,Overcoming Catastrophic Interference using Conceptor-Aided Backpropagation,"Catastrophic interference has been a major roadblock in the research of continual learning.Here we propose a variant of the back-propagation algorithm, ""Conceptor-Aided Backprop"", in which gradients are shielded by conceptors against degradation of previously learned tasks.Conceptors have their origin in reservoir computing, where they have been previously shown to overcome catastrophic forgetting.CAB extends these results to deep feedforward networks.On the disjoint and permuted MNIST tasks, CAB outperforms two other methods for coping with catastrophic interference that have recently been proposed.","We propose a variant of the backpropagation algorithm, in which gradients are shielded by conceptors against degradation of previously learned tasks.This paper applies the notion of conceptors, a form a regulariser, to prevent forgetting in continual learning in the training of neural networks on sequential tasks.Introduces a method for learning new tasks, without interfering previous tasks, using conceptors."
67,Why Do Neural Response Generation Models Prefer Universal Replies?,"Recent advances in neural Sequence-to-Sequence models reveal a purely data-driven approach to the response generation task.Despite its diverse variants and applications, the existing Seq2Seq models are prone to producing short and generic replies, which blocks such neural network architectures from being utilized in practical open-domain response generation tasks.In this research, we analyze this critical issue from the perspective of the optimization goal of models and the specific characteristics of human-to-human conversational corpora.Our analysis is conducted by decomposing the goal of Neural Response Generation into the optimizations of word selection and ordering.It can be derived from the decomposing that Seq2Seq based NRG models naturally tend to select common words to compose responses, and ignore the semantic of queries in word ordering.On the basis of the analysis, we propose a max-marginal ranking regularization term to avoid Seq2Seq models from producing the generic and uninformative responses.The empirical experiments on benchmarks with several metrics have validated our analysis and proposed methodology.",Analyze the reason for neural response generative models preferring universal replies; Propose a method to avoid it.Investigates the problem of universal replies plaguing the Seq2Seq neural generation modelsThe paper looks into improving the neural response generation task by deemphasizing the common responses using modification of the loss function and presentation the common/universal responses during the training phase.
68,code2seq: Generating Sequences from Structured Representations of Code,"The ability to generate natural language sequences from source code snippets has a variety of applications such as code summarization, documentation, and retrieval.Sequence-to-sequence models, adopted from neural machine translation, have achieved state-of-the-art performance on these tasks by treating source code as a sequence of tokens.We present code2seq: an alternative approach that leverages the syntactic structure of programming languages to better encode source code.Our model represents a code snippet as the set of compositional paths in its abstract syntax tree and uses attention to select the relevant paths while decoding.We demonstrate the effectiveness of our approach for two tasks, two programming languages, and four datasets of up to 16M examples.Our model significantly outperforms previous models that were specifically designed for programming languages, as well as general state-of-the-art NMT models.An interactive online demo of our model is available at http://code2seq.org.Our code, data and trained models are available at http://github.com/tech-srl/code2seq.",We leverage the syntactic structure of source code to generate natural language sequences.Presents a method for generating sequences from code by parsing and producing a syntax treeThis paper introduces an AST-based encoding for programming code and shows its effectiveness in the tasks of extreme code summarization and code captioning.This paper presents a new code-to-sequence model that leverages the syntactic structure of programming languages to encode source code snippets and then decode them to natural language
69,A Painless Attention Mechanism for Convolutional Neural Networks,"We propose a novel attention mechanism to enhance Convolutional Neural Networks for fine-grained recognition.The proposed mechanism reuses CNN feature activations to find the most informative parts of the image at different depths with the help of gating mechanisms and without part annotations.Thus, it can be used to augment any layer of a CNN to extract low- and high-level local information to be more discriminative.Differently, from other approaches, the mechanism we propose just needs a single pass through the input and it can be trained end-to-end through SGD.As a consequence, the proposed mechanism is modular, architecture-independent, easy to implement, and faster than iterative approaches.Experiments show that, when augmented with our approach, Wide Residual Networks systematically achieve superior performance on each of five different fine-grained recognition datasets: the Adience age and gender recognition benchmark, Caltech-UCSD Birds-200-2011, Stanford Dogs, Stanford Cars, and UEC Food-100, obtaining competitive and state-of-the-art scores.",We enhance CNNs with a novel attention mechanism for fine-grained recognition. Superior performance is obtained on 5 datasets.Describes a novel attentional mechanism applid to fine-grained recognition that consistently improves the recognition accuracy of the baselineThis paper proposes a feed-forward attention mechanism for fine-grained image classificationThis paper presents an interesting attention mechanism for fine-grained image classification.
70,Generative Adversarial Networks using Adaptive Convolution,"Most existing GANs architectures that generate images use transposed convolution or resize-convolution as their upsampling algorithm from lower to higher resolution feature maps in the generator.We argue that this kind of fixed operation is problematic for GANs to model objects that have very different visual appearances.We propose a novel adaptive convolution method that learns the upsampling algorithm based on the local context at each location to address this problem.We modify a baseline GANs architecture by replacing normal convolutions with adaptive convolutions in the generator.Experiments on CIFAR-10 dataset show that our modified models improve the baseline model by a large margin.Furthermore, our models achieve state-of-the-art performance on CIFAR-10 and STL-10 datasets in the unsupervised setting.","We replace normal convolutions with adaptive convolutions to improve GANs generator.Proposes to replace convolutions in the generator with an Adaptive Convolution Block that learns to generate convolution weigths adn biases of upsampling operations adaptively per pixel locationUses Adaptive Convolution in the context of GANs with a block called AdaConvBlock that replaces regular Convolution, this gives more local context per kernel weight so that it can generate locally flexible objects."
71,Large scale distributed neural network training through online distillation,"Techniques such as ensembling and distillation promise model quality improvements when paired with almost any base model.However, due to increased test-time cost and increased complexity of the training pipeline, these techniques are challenging to use in industrial settings.In this paper we explore a variant of distillation which is relatively straightforward to use as it does not require a complicated multi-stage setup or many new hyperparameters.Our first claim is that online distillation enables us to use extra parallelism to fit very large datasets about twice as fast.Crucially, we can still speed up training even after we have already reached the point at which additional parallelism provides no benefit for synchronous or asynchronous stochastic gradient descent.Two neural networks trained on disjoint subsets of the data can share knowledge by encouraging each model to agree with the predictions the other model would have made.These predictions can come from a stale version of the other model so they can be safely computed using weights that only rarely get transmitted.Our second claim is that online distillation is a cost-effective way to make the exact predictions of a model dramatically more reproducible.We support our claims using experiments on the Criteo Display Ad Challenge dataset, ImageNet, and the largest to-date dataset used for neural language modeling, containing tokens and based on the Common Crawl repository of web data.","We perform large scale experiments to show that a simple online variant of distillation can help us scale distributed neural network training to more machines.Proposes a method to scale distributed training beyond the current limits of mini-batch stochastic gradient descentProposal for an online distillation method called co-distillation, applied at scale, where two different models are trained to match predictions of the other model in addition to minimizing its own loss.Online distillation technique is introduced to accelerate traditional algorithms for large-scaled distributed neural network training"
72,Small Coresets to Represent Large Training Data for Support Vector Machines,"Support Vector Machines are one of the most popular algorithms for classification and regression analysis.Despite their popularity, even efficient implementations have proven to be computationally expensive to train at a large-scale, especially in streaming settings.In this paper, we propose a novel coreset construction algorithm for efficiently generating compact representations of massive data sets to speed up SVM training.A coreset is a weighted subset of the original data points such that SVMs trained on the coreset are provably competitive with those trained on the original data set.We provide both lower and upper bounds on the number of samples required to obtain accurate approximations to the SVM problem as a function of the complexity of the input data.Our analysis also establishes sufficient conditions on the existence of sufficiently compact and representative coresets for the SVM problem.We empirically evaluate the practical effectiveness of our algorithm against synthetic and real-world data sets.",We present an algorithm for speeding up SVM training on massive data sets by constructing compact representations that provide efficient and provably approximate inference.Studies the approach of coreset for SVM and aims at sampling a small set of weighted points such that the loss function over the points provably approximates that over the whole datasetThe paper suggests an importance sampling based Coreset construction to represent large training data for SVMs
73,Convergence rate of sign stochastic gradient descent for non-convex functions,"The sign stochastic gradient descent method utilizes only the sign of the stochastic gradient in its updates.Since signSGD carries out one-bit quantization of the gradients, it is extremely practical for distributed optimization where gradients need to be aggregated from different processors.For the first time, we establish convergence rates for signSGD on general non-convex functions under transparent conditions.We show that the rate of signSGD to reach first-order critical points matches that of SGD in terms of number of stochastic gradient calls, up to roughly a linear factor in the dimension.We carry out simple experiments to explore the behaviour of sign gradient descent close to saddle points and show that it often helps completely avoid them without using either stochasticity or curvature information.","We prove a non-convex convergence rate for the sign stochastic gradient method. The algorithm has links to algorithms like Adam and Rprop, as well as gradient quantisation schemes used in distributed machine learning.Provided a convergence analysis of Sign SGD algorithm for non-covex casesThe paper explores an algorithm that uses the sign of the gradients instead of actual gradients for training deep models"
74,Towards Transparent Neural Network Acceleration,"Deep learning has found numerous applications thanks to its versatility and accuracy on pattern recognition problems such as visual object detection.Learning and inference in deep neural networks, however, are memory and compute intensive and so improving efficiency is one of the major challenges for frameworks such as PyTorch, Tensorflow, and Caffe.While the efficiency problem can be partially addressed with specialized hardware and its corresponding proprietary libraries, we believe that neural network acceleration should be transparent to the user and should support all hardware platforms and deep learning libraries.To this end, we introduce a transparent middleware layer for neural network acceleration.The system is built around a compiler for deep learning, allowing one to combine device-specific libraries and custom optimizations while supporting numerous hardware devices.In contrast to other projects, we explicitly target the optimization of both prediction and training of neural networks.We present the current development status and some preliminary but encouraging results: on a standard x86 server, using CPUs our system achieves a 11.8x speed-up for inference and a 8.0x for batched-prediction; on GPUs we achieve a 1.7x and 2.3x speed-up respectively.","We introduce a transparent middleware for neural network acceleration, with own compiler engine, achieving up to 11.8x speed up on CPUs and 2.3x on GPUs.This paper proposes a transparent middleware layer for neural network acceleration and obtains some acceleration results on basic CPU and GPU architectures"
75,Rotation Equivariant Networks via Conic Convolution and the DFT,"Performance of neural networks can be significantly improved by encoding known invariance for particular tasks.Many image classification tasks, such as those related to cellular imaging, exhibit invariance to rotation.In particular, to aid convolutional neural networks in learning rotation invariance, we consider a simple, efficient conic convolutional scheme that encodes rotational equivariance, along with a method for integrating the magnitude response of the 2D-discrete-Fourier transform to encode global rotational invariance.We call our new method the Conic Convolution and DFT Network.We evaluated the efficacy of CFNet as compared to a standard CNN and group-equivariant CNN for several different image classification tasks and demonstrated improved performance, including classification accuracy, computational efficiency, and its robustness to hyperparameter selection.Taken together, we believe CFNet represents a new scheme that has the potential to improve many imaging analysis applications.","We propose conic convolution and the 2D-DFT to encode rotation equivariance into an neural network.In the context of image classification, the paper proposes a convolutional neural network architecture with rotation-equivariant feature maps that are eventually made rotation-invariant by using the magnitude of the 2D discrete Fourier transform (DFT).Authors provide a rotation invariant neural network via combining conic convolution and 2D-DFT"
76,Towards Metamerism via Foveated Style Transfer,"The problem of visual metamerism is defined as finding a family of perceptuallyindistinguishable, yet physically different images.In this paper, we propose ourNeuroFovea metamer model, a foveated generative model that is based on a mixtureof peripheral representations and style transfer forward-pass algorithms.Ourgradient-descent free model is parametrized by a foveated VGG19 encoder-decoderwhich allows us to encode images in high dimensional space and interpolatebetween the content and texture information with adaptive instance normalizationanywhere in the visual field.Our contributions include:1) A framework forcomputing metamers that resembles a noisy communication system via a foveatedfeed-forward encoder-decoder network – We observe that metamerism arises as abyproduct of noisy perturbations that partially lie in the perceptual null space;2)A perceptual optimization scheme as a solution to the hyperparametric nature ofour metamer model that requires tuning of the image-texture tradeoff coefficientseverywhere in the visual field which are a consequence of internal noise;3) AnABX psychophysical evaluation of our metamers where we also find that the rateof growth of the receptive fields in our model match V1 for reference metamersand V2 between synthesized samples.Our model also renders metamers at roughlya second, presenting a ×1000 speed-up compared to the previous work, which nowallows for tractable data-driven metamer experiments.","We introduce a novel feed-forward framework to generate visual metamersProposes a NeuroFovea model for generation of point-of-fixation metamers by using a style transfer approach via and Encoder-Decoder style architectureAn analysis of metamerism and a model capable of rapidly producing metamers of value for experimental psychophysics and other domains.The paper proposes a fast method for generating visual metamers – physically different images that cannot be told apart from an original – via foveated, fast, arbitrary style transfer"
77,On the Margin Theory of Feedforward Neural Networks,"Past works have shown that, somewhat surprisingly, over-parametrization can help generalization in neural networks.Towards explaining this phenomenon, we adopt a margin-based perspective.We establish:1) for multi-layer feedforward relu networks, the global minimizer of a weakly-regularized cross-entropy loss has the maximum normalized margin among all networks,2) as a result, increasing the over-parametrization improves the normalized margin and generalization error bounds for deep networks.In the case of two-layer networks, an infinite-width neural network enjoys the best generalization guarantees.The typical infinite feature methods are kernel methods; we compare the neural net margin with that of kernel methods and construct natural instances where kernel methods have much weaker generalization guarantees.We validate this gap between the two approaches empirically.Finally, this infinite-neuron viewpoint is also fruitful for analyzing optimization.We show that a perturbed gradient flow on infinite-size networks finds a global optimizer in polynomial time.","We show that training feedforward relu networks with a weak regularizer results in a maximum margin and analyze the implications of this result.Studies margin theory for neural sets and shows that max margin is monotonically increasing in size of the networkThis paper studies the implicit bias of minimizers of a regularized cross entropy loss of a two-layer network with ReLU activations, obtaining a generalization upper bound which does not increase with the network size."
78,Neural Networks for irregularly observed continuous-time Stochastic Processes,"Designing neural networks for continuous-time stochastic processes is challenging, especially when observations are made irregularly.In this article, we analyze neural networks from a frame theoretic perspective to identify the sufficient conditions that enable smoothly recoverable representations of signals in L^2.Moreover, we show that, under certain assumptions, these properties hold even when signals are irregularly observed.As we converge to the family of neural networks that satisfy these conditions, we show that we can optimize our convolution filters while constraining them so that they effectively compute a Discrete Wavelet Transform.Such a neural network can efficiently divide the time-axis of a signal into orthogonal sub-spaces of different temporal scale and localization.We evaluate the resulting neural network on an assortment of synthetic and real-world tasks: parsimonious auto-encoding, video classification, and financial forecasting.","Neural architectures providing representations of irregularly observed signals that provably enable signal reconstruction.Proves that convolutional neural networks with Leaky ReLU activation function are nonlinear frames, with similar results for non-uniformly sampled time-seriesThis article considers neural networks over time-series and show that the first convolutional filters can be chosen to represent a discrete wavelet transform."
79,Phrase-Based Attentions,"Most state-of-the-art neural machine translation systems, despite being differentin architectural skeletons, share an indispensablefeature: the Attention.However, most existing attention methods are token-basedand ignore the importance of phrasal alignments, the key ingredient for the successof phrase-based statistical machine translation.In this paper, we proposenovel phrase-based attention methods to model n-grams of tokens as attentionentities.We incorporate our phrase-based attentions into the recently proposedTransformer network, and demonstrate that our approach yields improvements of1.3 BLEU for English-to-German and 0.5 BLEU for German-to-English translationtasks, and 1.75 and 1.35 BLEU points in English-to-Russian and Russian-to-English translation taskson WMT newstest2014 using WMT’16 training data.","Phrase-based attention mechanisms to assign attention on phrases, achieving token-to-phrase, phrase-to-token, phrase-to-phrase attention alignments, in addition to existing token-to-token attentions.Paper presents an attention mechanism that computes a weighted sum over not only single tokens but ngrams(phrases)."
80,Reducing Overconfident Errors outside the Known Distribution,"Intuitively, unfamiliarity should lead to lack of confidence.In reality, current algorithms often make highly confident yet wrong predictions when faced with unexpected test samples from an unknown distribution different from training.Unlike domain adaptation methods, we cannot gather an ""unexpected dataset"" prior to test, and unlike novelty detection methods, a best-effort original task prediction is still expected.We compare a number of methods from related fields such as calibration and epistemic uncertainty modeling, as well as two proposed methods that reduce overconfident errors of samples from an unknown novel distribution without drastically increasing evaluation time: G-distillation, training an ensemble of classifiers and then distill into a single model using both labeled and unlabeled examples, or NCR, reducing prediction confidence based on its novelty detection score.Experimentally, we investigate the overconfidence problem and evaluate our solution by creating ""familiar"" and ""novel"" test splits, where ""familiar"" are identically distributed with training and ""novel"" are not.We discover that calibrating using temperature scaling on familiar data is the best single-model method for improving novel confidence, followed by our proposed methods.In addition, some methods' NLL performance are roughly equivalent to a regularly trained model with certain degree of smoothing.Calibrating can also reduce confident errors, for example, in gender recognition by 95% on demographic groups different from the training data.","Deep networks are more likely to be confidently wrong when testing on unexpected data. We propose an experimental methodology to study the problem, and two methods to reduce confident errors on unknown input distributions.Proposes two ideas for reducing overconfident wrong predictions: ""G-distillation"" of am ensemble with extra unsupervised data and Novelty Confidence Reduction using novelty detectorThe authors propose two methods for estimating classification confidence on novel unseen data distributions. The first idea is to use ensemble methods as the base approach to help identify uncertain cases and then use distillation methods to reduce the ensemble into a single model mimicking behavior of the ensemble. The second idea is to use a novelty detector classifier and weight the network output by the novelty score."
81,Neumann Optimizer: A Practical Optimization Algorithm for Deep Neural Networks,"Progress in deep learning is slowed by the days or weeks it takes to train large models.The natural solution of using more hardware is limited by diminishing returns, and leads to inefficient use of additional resources.In this paper, we present a large batch, stochastic optimization algorithm that is both faster than widely used algorithms for fixed amounts of computation, and also scales up substantially better as more computational resources become available.Our algorithm implicitly computes the inverse Hessian of each mini-batch to produce descent directions; we do so without either an explicit approximation to the Hessian or Hessian-vector products.We demonstrate the effectiveness of our algorithm by successfully training large ImageNet models with mini-batch sizes of up to 32000 with no loss in validation error relative to current baselines, and no increase in the total number of steps.At smaller mini-batch sizes, our optimizer improves the validation error in these models by 0.8-0.9%.Alternatively, we can trade off this accuracy to reduce the number of training steps needed by roughly 10-30%.Our work is practical and easily usable by others -- only one hyperparameter needs tuning, and furthermore, the algorithm is as computationally cheap as the commonly used Adam optimizer.",We describe a practical optimization algorithm for deep neural networks that works faster and generates better models compared to widely used algorithms.Proposes a new algorithm where they claim to use Hessian implicitly and are using a motivation from power-seriesPresents a new 2nd-order algorithm that implicitly uses curvature information and shows the intuition behind the approximation schemes in the algorithms and validates the heuristics in various experiments.
82,Heterogeneous Bitwidth Binarization in Convolutional Neural Networks,"Recent work has shown that performing inference with fast, very-low-bitwidth representations of values in models can yield surprisingly accurateresults.However, although 2-bit approximated networks have been shown tobe quite accurate, 1 bit approximations, which are twice as fast, have restrictivelylow accuracy.We propose a method to train models whose weights are a mixtureof bitwidths, that allows us to more finely tune the accuracy/speed trade-off.Wepresent the “middle-out” criterion for determining the bitwidth for each value, andshow how to integrate it into training models with a desired mixture of bitwidths.We evaluate several architectures and binarization techniques on the ImageNetdataset.We show that our heterogeneous bitwidth approximation achieves superlinearscaling of accuracy with bitwidth.Using an average of only 1.4 bits, we areable to outperform state-of-the-art 2-bit architectures.",We introduce fractional bitwidth approximation and show it has significant advantages.Suggests a method for varying the degree of quantization in a neural network during the forward propagation phaseMaintaining the accuracy of 2bits netword while using less than 2bits weights
83,Mean Replacement Pruning  ,"Pruning units in a deep network can help speed up inference and training as well as reduce the size of the model.We show that bias propagation is a pruning technique which consistently outperforms the common approach of merely removing units, regardless of the architecture and the dataset. We also show how a simple adaptation to an existing scoring function allows us to select the best units to prune. Finally, we show that the units selected by the best performing scoring functions are somewhat consistent over the course of training, implying the dead parts of the network appear during the stages of training.","Mean Replacement is an efficient method to improve the loss after pruning and Taylor approximation based scoring functions works better with absolute values. Proposes a simple improvement to methods for unit pruning using ""mean replacement""This paper presents a mean-replacement pruning strategy and utilizes the absolute-valued Taylor expansion as the scoring function for the pruning"
84,Preventing Posterior Collapse with delta-VAEs,"Due to the phenomenon of “posterior collapse,” current latent variable generative models pose a challenging design choice that either weakens the capacity of the decoder or requires altering the training objective.We develop an alternative that utilizes the most powerful generative models as decoders, optimize the variational lower bound, and ensures that the latent variables preserve and encode useful information.Our proposed δ-VAEs achieve this by constraining the variational family for the posterior to have a minimum distance to the prior.For sequential latent variable models, our approach resembles the classic representation learning approach of slow feature analysis.We demonstrate our method’s efficacy at modeling text on LM1B and modeling images: learning representations, improving sample quality, and achieving state of the art log-likelihood on CIFAR-10 and ImageNet 32 × 32.", Avoid posterior collapse by lower bounding the rate.Presents an approach to preventing posterior collapse in VAEs by limiting the family of the variational approximation to the posteriorThis paper introduces a constraint on the family of variational posteriors such that the KL term can be controlled to combat posterior collapse in deep generative models such as VAEs
85,On Batch Adaptive Training for Deep Learning: Lower Loss and Larger Step Size,"Mini-batch gradient descent and its variants are commonly used in deep learning.The principle of mini-batch gradient descent is to use noisy gradient calculated on a batch to estimate the real gradient, thus balancing the computation cost per iteration and the uncertainty of noisy gradient.However, its batch size is a fixed hyper-parameter requiring manual setting before training the neural network.Yin et al. proposed a batch adaptive stochastic gradient descent that can dynamically choose a proper batch size as learning proceeds.We extend the BA-SGD to momentum algorithm and evaluate both the BA-SGD and the batch adaptive momentum on two deep learning tasks from natural language processing to image classification.Experiments confirm that batch adaptive methods can achieve a lower loss compared with mini-batch methods after scanning the same epochs of data.Furthermore, our BA-Momentum is more robust against larger step sizes, in that it can dynamically enlarge the batch size to reduce the larger uncertainty brought by larger step sizes.We also identified an interesting phenomenon, batch size boom.The code implementing batch adaptive framework is now open source, applicable to any gradient-based optimization problems.","We developed a batch adaptive momentum that can achieve lower loss compared with mini-batch methods after scanning same epochs of data, and it is more robust against large step size.This paper addresses the problem of automatically tuning batch size during deep learning training, and claims to extend batch adaptive SGD to adaptive momentum and adopt the algorithms to complex neural networks problems.The paper proposes generalizing an algorithm which performs SGD with adaptive batch sizes by adding momentum to the utility function"
86,Adversarial Attacks on Graph Neural Networks via Meta Learning,"Deep learning models for graphs have advanced the state of the art on many tasks.Despite their recent success, little is known about their robustness.We investigate training time attacks on graph neural networks for node classification that perturb the discrete graph structure. Our core principle is to use meta-gradients to solve the bilevel problem underlying training-time attacks, essentially treating the graph as a hyperparameter to optimize.Our experiments show that small graph perturbations consistently lead to a strong decrease in performance for graph convolutional networks, and even transfer to unsupervised embeddings.Remarkably, the perturbations created by our algorithm can misguide the graph neural networks such that they perform worse than a simple baseline that ignores all relational information.Our attacks do not assume any knowledge about or access to the target classifiers.","We use meta-gradients to attack the training procedure of deep neural networks for graphs.Studies the problem of learning a better poisoned graph parameters that can maximize the loss of a graph neural network. An algorithm to alter graph structure by adding/deleting edges so as to degrade the global performance of node classification, and the idea to use meta-learning to solve the bilevel optimization problem."
87,Systematic Generalization: What Is Required and Can It Be Learned?,"Numerous models for grounded language understanding have been recently proposed, including generic models that can be easily adapted to any given task and intuitively appealing modular models that require background knowledge to be instantiated.We compare both types of models in how much they lend themselves to a particular form of systematic generalization.Using a synthetic VQA test, we evaluate which models are capable of reasoning about all possible object pairs after training on only a small subset of them.Our findings show that the generalization of modular models is much more systematic and that it is highly sensitive to the module layout, i.e. to how exactly the modules are connected.We furthermore investigate if modular models that generalize well could be made more end-to-end by learning their layout and parametrization.We find that end-to-end methods from prior work often learn inappropriate layouts or parametrizations that do not facilitate systematic generalization.Our results suggest that, in addition to modularity, systematic generalization in language understanding may require explicit regularizers or priors.","We show that modular structured models are the best in terms of systematic generalization and that their end-to-end versions don't generalize as well."", 'This paper evaluates systemic generalization between modular neural networks and otherwise generic models via introduction of a new, spatial reasoning datasetA targeted empirical evaluation of generalization in models for visual reasoning, focused on the problem of recognizing (object, relation, object) triples in synthetic scenes featuring letters and numbers."
88,Relational Forward Models for Multi-Agent Learning,"The behavioral dynamics of multi-agent systems have a rich and orderly structure, which can be leveraged to understand these systems, and to improve how artificial agents learn to operate in them.Here we introduce Relational Forward Models for multi-agent learning, networks that can learn to make accurate predictions of agents' future behavior in multi-agent environments."", ""Because these models operate on the discrete entities and relations present in the environment, they produce interpretable intermediate representations which offer insights into what drives agents' behavior, and what events mediate the intensity and valence of social interactions.Furthermore, we show that embedding RFM modules inside agents results in faster learning systems compared to non-augmented baselines.As more and more of the autonomous systems we develop and interact with become multi-agent in nature, developing richer analysis tools for characterizing how and why agents make decisions is increasingly necessary.Moreover, developing artificial agents that quickly and safely learn to coordinate with one another, and with humans in shared environments, is crucial.","Relational Forward Models for multi-agent learning make accurate predictions of agents' future behavior, they produce intepretable representations and can be used inside agents."", 'A way of reducing variance in model free learning by having an explicit model, that uses a graph conv net-like architecture, of actions that other agents will take. Predicting multi-agent behavior using a relational forward model with a recurrent component, outperforming two baselines and two ablations"
89,The Implicit Bias of Gradient Descent on Separable Data,"We show that gradient descent on an unregularized logistic regressionproblem, for almost all separable datasets, converges to the same direction as the max-margin solution.The result generalizes also to other monotone decreasing loss functions with an infimum at infinity, and we also discuss a multi-class generalizations to the cross entropy loss.Furthermore,we show this convergence is very slow, and only logarithmic in theconvergence of the loss itself.This can help explain the benefitof continuing to optimize the logistic or cross-entropy loss evenafter the training error is zero and the training loss is extremelysmall, and, as we show, even if the validation loss increases.Ourmethodology can also aid in understanding implicit regularizationin more complex models and with other optimization methods.","The normalized solution of gradient descent on logistic regression (or a similarly decaying loss) slowly converges to the L2 max margin solution on separable data.The paper offers a formal proof that gradient descent on the logistic loss converges very slowly to the hard SVM solution in the case where the data are linearly separable. This paper focuses on characterising the behaviour of log loss minimisation on linearly separable data, and shows that log-loss, minimised with gradient descent, leads to convergence to the max-margin solution."
90,Learning Robust Representations by Projecting Superficial Statistics Out,"Despite impressive performance as evaluated on i.i.d. holdout data, deep neural networks depend heavily on superficial statistics of the training data and are liable to break under distribution shift.For example, subtle changes to the background or texture of an image can break a seemingly powerful classifier.Building on previous work on domain generalization, we hope to produce a classifier that will generalize to previously unseen domains, even when domain identifiers are not available during training.This setting is challenging because the model may extract many distribution-specific signals together with distribution-agnostic signals.To overcome this challenge, we incorporate the gray-level co-occurrence matrix to extract patterns that our prior knowledge suggests are superficial: they are sensitive to the texture but unable to capture the gestalt of an image.Then we introduce two techniques for improving our networks' out-of-sample performance.The first method is built on the reverse gradient method that pushes our model to learn representations from which the GLCM representation is not predictable.The second method is built on the independence introduced by projecting the model's representation onto the subspace orthogonal to GLCM representation's.We test our method on the battery of standard domain generalization data sets and, interestingly, achieve comparable or better performance as compared to other domain generalization methods that explicitly require samples from the target distribution for training.","Building on previous work on domain generalization, we hope to produce a classifier that will generalize to previously unseen domains, even when domain identifiers are not available during training.A domain generalization approach to reveal semantic information based on a linear projection scheme from CNN and NGLCM output layers.The paper proposes an unsupervised approach to identify image features that are not meaningful for image classification tasks"
91,CAMOU: Learning Physical Vehicle Camouflages to Adversarially Attack Detectors in the Wild,"In this paper, we conduct an intriguing experimental study about the physical adversarial attack on object detectors in the wild.In particular, we learn a camouflage pattern to hide vehicles from being detected by state-of-the-art convolutional neural network based detectors.Our approach alternates between two threads.In the first, we train a neural approximation function to imitate how a simulator applies a camouflage to vehicles and how a vehicle detector performs given images of the camouflaged vehicles.In the second, we minimize the approximated detection score by searching for the optimal camouflage.Experiments show that the learned camouflage can not only hide a vehicle from the image-based detectors under many test cases but also generalizes to different environments, vehicles, and object detectors.","We propose a method to learn physical vehicle camouflage to adversarially attack object detectors in the wild. We find our camouflage effective and transferable.The authors investigate the problem of learning a camouflage pattern which, when applied to a simulated vehicle, will prevent an object detector from detecting it.This paper targets adversarial learning for interfering car detection by learning camouflage patterns"
92,Interpretable Classification via Supervised Variational Autoencoders and Differentiable Decision Trees,"As deep learning-based classifiers are increasingly adopted in real-world applications, the importance of understanding how a particular label is chosen grows.Single decision trees are an example of a simple, interpretable classifier, but are unsuitable for use with complex, high-dimensional data.On the other hand, the variational autoencoder is designed to learn a factored, low-dimensional representation of data, but typically encodes high-likelihood data in an intrinsically non-separable way. We introduce the differentiable decision tree as a modular component of deep networks and a simple, differentiable loss function that allows for end-to-end optimization of a deep network to compress high-dimensional data for classification by a single decision tree. We also explore the power of labeled data in a supervised VAE with a Gaussian mixture prior, which leverages label information to produce a high-quality generative model with improved bounds on log-likelihood. We combine the SVAE with the DDT to get our classifier+VAE, which is competitive in both classification error and log-likelihood, despite optimizing both simultaneously and using a very simple encoder/decoder architecture.","We combine differentiable decision trees with supervised variational autoencoders to enhance interpretability of classification. This paper proposes a hybrid model of a variational autoencoder composed with a differentiable decision tree, and an accompanying training scheme, with experiments demonstrating tree classification performance, neg. log likelihood performance, and latent space interpretability.The paper tries to build an interpretable and accurate classifier via stacking a supervised VAE and a differentiable decision tree"
93,Regularized Learning for  Domain Adaptation under Label Shifts,"We propose Regularized Learning under Label shifts, a principled and a practical domain-adaptation algorithm to correct for shifts in the label distribution between a source and a target domain.We first estimate importance weights using labeled source data and unlabeled target data, and then train a classifier on the weighted source samples.We derive a generalization bound for the classifier on the target domain which is independent of the data dimensions, and instead only depends on the complexity of the function class.To the best of our knowledge, this is the first generalization bound for the label-shift problem where the labels in the target domain are not available.Based on this bound, we propose a regularized estimator for the small-sample regime which accounts for the uncertainty in the estimated weights.Experiments on the CIFAR-10 and MNIST datasets show that RLLS improves classification accuracy, especially in the low sample and large-shift regimes, compared to previous methods.","A practical and provably guaranteed approach for training efficiently classifiers in the presence of label shifts between Source and Target data setsThe authors propose a new algorithm for improving the stability of class importance weighting estimation procedure with a two-step procedure.The authors consider the problem of learning under label shifts, where label proportions differ while conditionals are equal, and propose an improved estimator with regularization."
94,Bayesian Embeddings for Long-Tailed Datasets,"The statistics of the real visual world presents a long-tailed distribution: a few classes have significantly more training instances than the remaining classes in a dataset.This is because the real visual world has a few classes that are common while others are rare.Unfortunately, the performance of a convolutional neural network is typically unsatisfactory when trained using a long-tailed dataset.To alleviate this issue, we propose a method that discriminatively learns an embedding in which a simple Bayesian classifier can balance the class-priors to generalize well for rare classes.To this end, the proposed approach uses a Gaussian mixture model to factor out class-likelihoods and class-priors in a long-tailed dataset.The proposed method is simple and easy-to-implement in existing deep learning frameworks.Experiments on publicly available datasets show that the proposed approach improves the performance on classes with few training instances, while maintaining a comparable performance to the state-of-the-art on classes with abundant training examples.","Approach to improve classification accuracy on classes in the tail.The main goal of this paper is to learn a ConvNet classifier which performs better for classes in the tail of the class occurrence distribution.Proposal for a Bayesian framework with a Gaussian mixture model to address an issue in classification applications, that the number of training data from different classes is unbalanced."
95,Visualizing and Discovering Behavioural Weaknesses in Deep Reinforcement Learning,"As deep reinforcement learning is being applied to more and more tasks, there is a growing need to better understand and probe the learned agents.Visualizing and understanding the decision making process can be very valuable to comprehend and identify problems in the learned behavior.However, this topic has been relatively under-explored in the reinforcement learning community.In this work we present a method for synthesizing states of interest for a trained agent.Such states could be situations in which specific actions are necessary.Further, critical states in which a very high or a very low reward can be achieved are often interesting to understand the situational awareness of the system.To this end, we learn a generative model over the state space of the environment and use its latent space to optimize a target function for the state of interest.In our experiments we show that this method can generate insightful visualizations for a variety of environments and reinforcement learning methods.We explore these issues in the standard Atari benchmark games as well as in an autonomous driving simulator.Based on the efficiency with which we have been able to identify significant decision scenarios with this technique, we believe this general approach could serve as an important tool for AI safety applications.",We present a method to synthesize states of interest for reinforcement learning agents in order to analyze their behavior. This paper proposes a generative model of visual observations in RL that is capable of generating observations of interests.An approach for visualizing states of interest that involves a variational autoencoder that learns to reconstruct state space and an optimization step that finds conditioning parameters to generate synthetic images.
96,Knows When it Doesn’t Know: Deep Abstaining Classifiers,"We introduce the deep abstaining classifier -- a deep neural network trained with a novel loss function that provides an abstention option during training.This allows the DNN to abstain on confusing or difficult-to-learn examples while improving performance on the non-abstained samples.We show that such deep abstaining classifiers can: learn representations for structured noise -- where noisy training labels or confusing examples are correlated with underlying features -- and then learn to abstain based on such features; enable robust learning in the presence of arbitrary or unstructured noise by identifying noisy samples; and be used as an effective out-of-category detector that learns to reliably abstain when presented with samples from unknown classes.We provide analytical results on loss function behavior that enable automatic tuning of accuracy and coverage, and demonstrate the utility of the deep abstaining classifier using multiple image benchmarks, Results indicate significant improvement in learning in the presence of label noise.","A deep abstaining neural network trained with a novel loss function that learns representations for when to abstain enabling robust learning in the presence of different types of noise.A new loss function for training a deep neural network which can abstain, with performance looked at from angles in existence of structured noise, in existence of unstructured noise, and open world detection.This manuscript introduces deep abstaining classifiers which modifies the multiclass cross-entropy loss with an abstention loss, which is then applied to perturbed image classification tasks"
97,HR-TD: A Regularized TD Method to Avoid Over-Generalization,"Temporal Difference learning with function approximation has been widely used recently and has led to several successful results. However, compared with the original tabular-based methods, one major drawback of temporal difference learning with neural networks and other function approximators is that they tend to over-generalize across temporally successive states, resulting in slow convergence and even instability.In this work, we propose a novel TD learning method, Hadamard product Regularized TD, that reduces over-generalization and thus leads to faster convergence.This approach can be easily applied to both linear and nonlinear function approximators.HR-TD is evaluated on several linear and nonlinear benchmark domains, where we show improvement in learning behavior and performance.","A regularization technique for TD learning that avoids temporal over-generalization, especially in Deep NetworksA variation on temporal difference learning for the function approximation case that attempts to resolve the issue of over-generalization across temporally-successive states.The paper introduces HR-TD, a variation of the TD(0) algorithm, meant to improve the over-generalization problem in conventional TD"
98,Spherical CNNs on Unstructured Grids,"We present an efficient convolution kernel for Convolutional Neural Networks on unstructured grids using parameterized differential operators while focusing on spherical signals such as panorama images or planetary signals.To this end, we replace conventional convolution kernels with linear combinations of differential operators that are weighted by learnable parameters.Differential operators can be efficiently estimated on unstructured grids using one-ring neighbors, and learnable parameters can be optimized through standard back-propagation.As a result, we obtain extremely efficient neural networks that match or outperform state-of-the-art network architectures in terms of performance but with a significantly lower number of network parameters.We evaluate our algorithm in an extensive series of experiments on a variety of computer vision and climate science tasks, including shape classification, climate pattern segmentation, and omnidirectional image semantic segmentation.Overall, we present a novel CNN approach on unstructured grids using parameterized differential operators for spherical signals, and we show that our unique kernel parameterization allows our model to achieve the same or higher accuracy with significantly fewer network parameters.","We present a new CNN kernel for unstructured grids for spherical signals, and show significant accuracy and parameter efficiency gain on tasks such as 3D classfication and omnidirectional image segmentation.An efficient method enabling deep learning on spherical data that reaches competitive/state-of-the-art numbers with much less parameters than popular approaches.The paper proposes a novel convolutional kernel for CNN on the unstructured grids and formulates the convolution by a linear combination of differential operators."
99,Time-Agnostic Prediction: Predicting Predictable Video Frames,"Prediction is arguably one of the most basic functions of an intelligent system.In general, the problem of predicting events in the future or between two waypoints is exceedingly difficult.However, most phenomena naturally pass through relatively predictable bottlenecks while we cannot predict the precise trajectory of a robot arm between being at rest and holding an object up, we can be certain that it must have picked the object up.To exploit this, we decouple visual prediction from a rigid notion of time.While conventional approaches predict frames at regularly spaced temporal intervals, our time-agnostic predictors are not tied to specific times so that they may instead discover predictable ""bottleneck"" frames no matter when they occur.We evaluate our approach for future and intermediate frame prediction across three robotic manipulation tasks.Our predictions are not only of higher visual quality, but also correspond to coherent semantic subgoals in temporally extended tasks.","In visual prediction tasks, letting your predictive model choose which times to predict does two things: (i) improves prediction quality, and (ii) leads to semantically coherent ""bottleneck state"" predictions, which are useful for planning.A method on prediction of frames in a video, the approach including that target prediction is floating, resolved by a minimum on the error of prediction.Reformulates the task of video prediction/interpolation so that a predictor is not forced to generate frames at fixed time intervals, but instead is trained to generate frames that happen at any point in the future."
100,Predicting Floor-Level for 911 Calls with Neural Networks and Smartphone Sensor Data,"In cities with tall buildings, emergency responders need an accurate floor level location to find 911 callers quickly.We introduce a system to estimate a victim's floor level via their mobile device's sensor data in a two-step process.First, we train a neural network to determine when a smartphone enters or exits a building via GPS signal changes.Second, we use a barometer equipped smartphone to measure the change in barometric pressure from the entrance of the building to the victim's indoor location.Unlike impractical previous approaches, our system is the first that does not require the use of beacons, prior knowledge of the building infrastructure, or knowledge of user behavior.We demonstrate real-world feasibility through 63 experiments across five different tall buildings throughout New York City where our system predicted the correct floor level with 100% accuracy.","We used an LSTM to detect when a smartphone walks into a building. Then we predict the device's floor level using data from sensors aboard the smartphone."", ""The paper introduces a system to estimate a floor-level via their mobile device's sensor data using an LSTM and changes in barometric pressure"", 'Proposal for a two-step method to determine which floor a mobile phone is on inside a tall building."
101,ACTRCE: Augmenting Experience via Teacher’s Advice,"Sparse reward is one of the most challenging problems in reinforcement learning.Hindsight Experience Replay attempts to address this issue by converting a failure experience to a successful one by relabeling the goals.Despite its effectiveness, HER has limited applicability because it lacks a compact and universal goal representation.We present Augmenting experienCe via TeacheR's adviCE, an efficient reinforcement learning technique that extends the HER framework using natural language as the goal representation.We first analyze the differences among goal representation, and show that ACTRCE can efficiently solve difficult reinforcement learning problems in challenging 3D navigation tasks, whereas HER with non-language goal representation failed to learn.We also show that with language goal representations, the agent can generalize to unseen instructions, and even generalize to instructions with unseen lexicons.We further demonstrate it is crucial to use hindsight advice to solve challenging tasks, but we also found that little amount of hindsight advice is sufficient for the learning to take off, showing the practical aspect of the method.","Combine language goal representation with hindsight experience replays.This paper considers the assumption implicit in hindsight experience replay, that there is access to a mapping from states to goals, and proposes a natural language goal representation.This submission uses Hindsight Experience Replay framework with natural language goals to improve the sample-efficiency of instruction-following models."
102,Efficient Codebook and Factorization for Second Order Representation Learning,"Learning rich and compact representations is an open topic in many fields such as word embedding, visual question-answering, object recognition or image retrieval.Although deep neural networks have made a major breakthrough during the last few years by providing hierarchical, semantic and abstract representations for all of these tasks, these representations are not necessary as rich as needed nor as compact as expected.Models using higher order statistics, such as bilinear pooling, provide richer representations at the cost of higher dimensional features.Factorization schemes have been proposed but without being able to reach the original compactness of first order models, or at a heavy loss in performances.This paper addresses these two points by extending factorization schemes to codebook strategies, allowing compact representations with the same dimensionality as first order representations, but with second order performances.Moreover, we extend this framework with a joint codebook and factorization scheme, granting a reduction both in terms of parameters and computation cost.This formulation leads to state-of-the-art results and compact second-order models with few additional parameters and intermediate representations with a dimension similar to that of first-order statistics.","We propose a joint codebook and factorization scheme to improve second order pooling.This paper presents a way to combine existing factorized second order representations with a codebook style hard assignment.Proposal for a novel bilinear representation based on a codebook model, and an efficient formulation in which codebook-based projections are factorized via shared projection to further reduce parameter size."
103,Semi-supervised Ensemble Learning with Weak Supervision for Biomedical Relationship Extraction,"Natural language understanding research has recently shifted towards complex Machine Learning and Deep Learning algorithms.Such models often outperform their simpler counterparts significantly.However, their performance relies on the availability of large amounts of labeled data, which are rarely available.To tackle this problem, we propose a methodology for extending training datasets to arbitrarily big sizes and training complex, data-hungry models using weak supervision.We apply this methodology on biomedical relation extraction, a task where training datasets are excessively time-consuming and expensive to create, yet has a major impact on downstream applications such as drug discovery.We demonstrate in two small-scale controlled experiments that our method consistently enhances the performance of an LSTM network, with performance improvements comparable to hand-labeled training data.Finally, we discuss the optimal setting for applying weak supervision using this methodology.","We propose and apply a meta-learning methodology based on Weak Supervision, for combining Semi-Supervised and Ensemble Learning on the task of Biomedical Relationship Extraction.A semi-supervised method for relation classification, which trains multiple base learners using a small labeled dataset and applies some of them to annotate unlabeled examples for semi-supervised learning.This paper addresses the problem of generating training data for biological relation extraction, and uses predictions from data labeled by weak classifiers as additional training data for a meta learning algorithm.This paper proposes a combination of semi-supervised learning and ensemble learning for information extraction, with experiments conducted on a biomedical relation extraction task"
104,Contextual Explanation Networks,"We introduce contextual explanation networks a class of models that learn to predict by generating and leveraging intermediate explanations.CENs are deep networks that generate parameters for context-specific probabilistic graphical models which are further used for prediction and play the role of explanations.Contrary to the existing post-hoc model-explanation tools, CENs learn to predict and to explain jointly.Our approach offers two major advantages: for each prediction, valid instance-specific explanations are generated with no computational overhead and prediction via explanation acts as a regularization and boosts performance in low-resource settings.We prove that local approximations to the decision boundary of our networks are consistent with the generated explanations.Our results on image and text classification and survival analysis tasks demonstrate that CENs are competitive with the state-of-the-art while offering additional insights behind each prediction, valuable for decision support.","A class of networks that generate simple models on the fly (called explanations) that act as a regularizer and enable consistent model diagnostics and interpretability.The authors claim that the previous art directly integrate neural networks into the graphical models as components, which renders the models uninterpretable.Proposal for a combination of neural nets and graphical models by using a deep neural net to predict the parameters of a graphical model."
105,Deterministic Policy Imitation Gradient Algorithm,"The goal of imitation learning is to enable a learner to imitate an expert’s behavior given the expert’s demonstrations.Recently, generative adversarial imitation learning has successfully achieved it even on complex continuous control tasks.However, GAIL requires a huge number of interactions with environment during training.We believe that IL algorithm could be more applicable to the real-world environments if the number of interactions could be reduced.To this end, we propose a model free, off-policy IL algorithm for continuous control.The keys of our algorithm are two folds:1) adopting deterministic policy that allows us to derive a novel type of policy gradient which we call deterministic policy imitation gradient,2) introducing a function which we call state screening function to avoid noisy policy updates with states that are not typical of those appeared on the expert’s demonstrations.Experimental results show that our algorithm can achieve the goal of IL with at least tens of times less interactions than GAIL on a variety of continuous control tasks.","We propose a model free imitation learning algorithm that is able to reduce number of interactions with environment in comparison with state-of-the-art imitation learning algorithm namely GAIL.Proposes to extend the determinist policy gradient algorithm to learn from demonstrations, while combined with a type of density estimation of the expert.This paper considers the problem of model-free imitation learning and proposes an extension of the generative adversarial imitation learning algorithm by replacing the stochastic policy of the learner with a deterministic one.The paper combines IRL, adversarial training, and ideas from deterministic policy gradients with the goal of decreasng sample complexity"
106,Topology Adaptive Graph Convolutional  Networks,"Convolution acts as a local feature extractor in convolutional neural networks.However, the convolution operation is not applicable when the input data is supported on an irregular graph such as with social networks, citation networks, or knowledge graphs.This paper proposes the topology adaptive graph convolutional network, a novel graph convolutional network that generalizes CNN architectures to graph-structured data and provides a systematic way to design a set of fixed-size learnable filters to perform convolutions on graphs.The topologies of these filters are adaptive to the topology of the graph when they scan the graph to perform convolution, replacing the square filter for the grid-structured data in traditional CNNs.The outputs are the weighted sum of these filters’ outputs, extraction of both vertex features and strength of correlation between vertices.Itcan be used with both directed and undirected graphs.The proposed TAGCN not only inherits the properties of convolutions in CNN for grid-structured data, but it is also consistent with convolution as defined in graph signal processing.Further, as no approximation to the convolution is needed, TAGCN exhibits better performance than existing graph-convolution-approximation methods on a numberof data sets.As only the polynomials of degree two of the adjacency matrix are used, TAGCN is also computationally simpler than other recent methods.","Low computational complexity graph CNN (without approximation) with better classification accuracyProposes a new CNN approach to graph classification using a filter based on outgoing walks of increasing length to incorporate information from more distant vertices in one propagation step.Proposal for a new neural network architecture for semi-supervised graph classification, building upon graph polynomial filters and utilizing them on successive neural network layers with ReLU activation functions.The paper introduces Topology Adaptive GCN to generalize convolutional networks to graph-structured data"
107,An Empirical Study of Example Forgetting during Deep Neural Network Learning,"Inspired by the phenomenon of catastrophic forgetting, we investigate the learning dynamics of neural networks as they train on single classification tasks.Our goal is to understand whether a related phenomenon occurs when data does not undergo a clear distributional shift.We define a forgetting event'' to have occurred when an individual training example transitions from being classified correctly to incorrectly over the course of learning.Across several benchmark data sets, we find that: certain examples are forgotten with high frequency, and some not at all;', "" a data set'sforgettable examples generalize across neural architectures; and"", ' based on forgetting dynamics, a significant fraction of examples can be omitted from the training data set while still maintaining state-of-the-art generalization performance.","We show that catastrophic forgetting occurs within what is considered to be a single task and find that examples that are not prone to forgetting can be removed from the training set without loss of generalization.Studies the forgetting behavior of training examples during SGD, and shows there exist ""support examples"" in neural network training across different network architectures.This paper analyzes the extent to which networks learn to correctly classify specific examples and then forget these examples over the course of training.The paper studies whether some examples in training neural networks are harder to learn than others. Such examples are forgotten and relearned multiple times through learning."
108,Object-Contrastive Networks: Unsupervised Object Representations,"Discovering objects and their attributes is of great importance for autonomous agents to effectively operate in human environments.This task is particularly challenging due to the ubiquitousness of objects and all their nuances in perceptual and semantic detail.In this paper we present an unsupervised approach for learning disentangled representations of objects entirely from unlabeled monocular videos.These continuous representations are not biased by or limited by a discrete set of labels determined by human labelers.The proposed representation is trained with a metric learning loss, where objects with homogeneous features are pushed together, while those with heterogeneous features are pulled apart.We show these unsupervised embeddings allow to discover object attributes and can enable robots to self-supervise in previously unseen environments.We quantitatively evaluate performance on a large-scale synthetic dataset with 12k object models, as well as on a real dataset collected by a robot and show that our unsupervised object understanding generalizes to previously unseen objects.Specifically, we demonstrate the effectiveness of our approach on robotic manipulation tasks, such as pointing at and grasping of objects.An interesting and perhaps surprising finding in this approach is that given a limited set of objects, object correspondences will naturally emerge when using metric learning without requiring explicit positive pairs.","An unsupervised approach for learning disentangled representations of objects entirely from unlabeled monocular videos.Designs a feature representation from video sequences captured from a scene from different view points.Proposal for an unsupervised representation learning method for visual inputs that incorporates a metric learning approach pulling nearest neighbor pairs of image patches close in embedding space while pushing apart other pairs.This paper explores self-supervised learning of object representations, with the main idea to encourage objects with similar features to get further ‘attracted’ to each other."
109,Quantile Regression Reinforcement Learning with State Aligned Vector Rewards,"Learning from a scalar reward in continuous action space environments is difficult and often requires millions if not billions of interactions. We introduce state aligned vector rewards, which are easily defined in metric state spaces and allow our deep reinforcement learning agent to tackle the curse of dimensionality. Our agent learns to map from action distributions to state change distributions implicitly defined in a quantile function neural network. We further introduce a new reinforcement learning technique inspired by quantile regression which does not limit agents to explicitly parameterized action distributions. Our results in high dimensional state spaces show that training with vector rewards allows our agent to learn multiple times faster than an agent training with scalar rewards.","We train with state aligned vector rewards an agent predicting state changes from action distributions, using a new reinforcement learning technique inspired by quantile regression.Presents algorithm that aims to speed up reinforcement learning in situations where the reward is aligned with the state space. This paper addresses RL in the continuous action space, using a re-parametrised policy and a novel vector-based training objective.This work proposes to mix distributional RL with a net in charge of modeling the evolution of the world in terms of quantiles, claiming improvements in sample efficiency."
110,Siamese Survival Analysis with Competing Risks,"Survival Analysis in the presence of multiple possible adverse events, i.e., competing risks, is a challenging, yet very important problem in medicine, finance, manufacturing, etc.Extending classical survival analysis to competing risks is not trivial since only one event is observed and hence, the incidence of an event of interest is often obscured by other related competing events.This leads to the nonidentifiability of the event times’ distribution parameters, which makes the problem significantly more challenging.In this work we introduce Siamese Survival Prognosis Network, a novel Siamese Deep Neural Network architecture that is able to effectively learn from data in the presence of multiple adverse events.The Siamese Survival Network is especially crafted to issue pairwise concordant time-dependent risks, in which longer event times are assigned lower risks.Furthermore, our architecture is able to directly optimize an approximation to the C-discrimination index, rather than relying on well-known metrics of cross-entropy etc., and which are not able to capture the unique requirements of survival analysis with competing risks.Our results show consistent performance improvements on a number of publicly available medical datasets over both statistical and deep learning state-of-the-art methods.",In this work we introduce a novel Siamese Deep Neural Network architecture that is able to effectively learn from data in the presence of multiple adverse events.This paper introduces siamese neural networks to the competing risks framework by optimizing for the c-index directlyThe authors address issues of estimating risk in a survival analysis setting with competing risks and propose directly optimizing the time-dependent discrimination index using a siamese survival network
111,Pointing Out SQL Queries From Text,"The digitization of data has resulted in making datasets available to millions of users in the form of relational databases and spreadsheet tables.However, a majority of these users come from diverse backgrounds and lack the programming expertise to query and analyze such tables.We present a system that allows for querying data tables using natural language questions, where the system translates the question into an executable SQL query.We use a deep sequence to sequence model in wich the decoder uses a simple type system of SQL expressions to structure the output prediction.Based on the type, the decoder either copies an output token from the input question using an attention-based copying mechanism or generates it from a fixed vocabulary.We also introduce a value-based loss function that transforms a distribution over locations to copy from into a distribution over the set of input tokens to improve training of our model.We evaluate our model on the recently released WikiSQL dataset and show that our model trained using only supervised learning significantly outperforms the current state-of-the-art Seq2SQL model that uses reinforcement learning.",We present a type-based pointer network model together with a value-based loss method to effectively train a neural model to translate natural language to SQL.The paper claims to develop a novel method to map natural language queries to SQL by using a grammar to guide decoding and using a new loss function for pointer / copy mechanism
112,ARM: Augment-REINFORCE-Merge Gradient for Stochastic Binary Networks,"To backpropagate the gradients through stochastic binary layers, we propose the augment-REINFORCE-merge estimator that is unbiased, exhibits low variance, and has low computational complexity.Exploiting variable augmentation, REINFORCE, and reparameterization, the ARM estimator achieves adaptive variance reduction for Monte Carlo integration by merging two expectations via common random numbers.The variance-reduction mechanism of the ARM estimator can also be attributed to either antithetic sampling in an augmented space, or the use of an optimal anti-symmetric ""self-control"" baseline function together with the REINFORCE estimator in that augmented space.Experimental results show the ARM estimator provides state-of-the-art performance in auto-encoding variational inference and maximum likelihood estimation, for discrete latent variable models with one or multiple stochastic binary layers.Python code for reproducible research is publicly available.",An unbiased and low-variance gradient estimator for discrete latent variable modelsProposes a new variance-reduction technique to use when computing an expected loss gradient where the expectation is with respect to independent binary random variables.An algorithm combining Rao-Blackwellization and common random numbers for lowering the variance of the score-function gradient estimator in the special case of stochastic binary networksAn unbiased and low variance augment-REINFORCE-merge (ARM) estimator for calculating and backpropagating gradients in binary neural networks
113,Local SGD Converges Fast and Communicates Little,"Mini-batch stochastic gradient descent is state of the art in large scale distributed training.The scheme can reach a linear speed-up with respect to the number of workers, but this is rarely seen in practice as the scheme often suffers from large network delays and bandwidth limits.To overcome this communication bottleneck recent works propose to reduce the communication frequency.An algorithm of this type is local SGD that runs SGD independently in parallel on different workers and averages the sequences only once in a while.This scheme shows promising results in practice, but eluded thorough theoretical analysis. We prove concise convergence rates for local SGD on convex problems and show that it converges at the same rate as mini-batch SGD in terms of number of evaluated gradients, that is, the scheme achieves linear speed-up in the number of workers and mini-batch size.The number of communication rounds can be reduced up to a factor of T^ where T denotes the number of total steps compared to mini-batch SGD.This also holds for asynchronous implementations.Local SGD can also be used for large scale training of deep learning models.The results shown here aim serving as a guideline to further explore the theoretical and practical aspects of local SGD in these applications.","We prove that parallel local SGD achieves linear speedup with much lesser communication than parallel mini-batch SGD.Provides a convergence proof for local SGD, and proves that local SGD can provide the same speedup gains as minibatch, but may be able to communicate significantly less.This paper presents an analysis of local SGD and bounds on how frequent the estimators obtained by running SGD required to be averaged in order to yield linear parallelization speedups.The authors analyze the local SGD algorithm, where parallel chains of SGD are run, and the iterates are occasionally synchronized across machines by averaging"
114,Learning Information Propagation in the Dynamical Systems via Information Bottleneck Hierarchy,"Extracting relevant information, causally inferring and predicting the future states with high accuracy is a crucial task for modeling complex systems.The endeavor to address these tasks is made even more challenging when we have to deal with high-dimensional heterogeneous data streams.Such data streams often have higher-order inter-dependencies across spatial and temporal dimensions.We propose to perform a soft-clustering of the data and learn its dynamics to produce a compact dynamical model while still ensuring the original objectives of causal inference and accurate predictions.To efficiently and rigorously process the dynamics of soft-clustering, we advocate for an information theory inspired approach that incorporates stochastic calculus and seeks to determine a trade-off between the predictive accuracy and compactness of the mathematical representation.We cast the model construction as a maximization of the compression of the state variables such that the predictive ability and causal interdependence constraints between the original data streams and the compact model are closely bounded.We provide theoretical guarantees concerning the convergence of the proposed learning algorithm.To further test the proposed framework, we consider a high-dimensional Gaussian case study and describe an iterative scheme for updating the new model parameters.Using numerical experiments, we demonstrate the benefits on compression and prediction accuracy for a class of dynamical systems.Finally, we apply the proposed algorithm to the real-world dataset of multimodal sentiment intensity and show improvements in prediction with reduced dimensions.",Compact perception of dynamical processStudies the problem of compactly representing the model of a complex dynamic system while preserving information by using an information bottleneck method.This paper studied the Gaussian linear dynamic and proposed an algorithm for computing the Information Bottleneck Hierarchy (IBH).
115,Dense Recurrent Neural Network with Attention Gate,"We propose the dense RNN, which has the fully connections from each hidden state to multiple preceding hidden states of all layers directly.As the density of the connection increases, the number of paths through which the gradient flows can be increased.It increases the magnitude of gradients, which help to prevent the vanishing gradient problem in time.Larger gradients, however, can also cause exploding gradient problem.To complement the trade-off between two problems, we propose an attention gate, which controls the amounts of gradient flows.We describe the relation between the attention gate and the gradient flows by approximation.The experiment on the language modeling using Penn Treebank corpus shows dense connections with the attention gate improve the model’s performance.","Dense RNN that has fully connections from each hidden state to multiple preceding hidden states of all layers directly.Proposes a new RNN architecture that models long-term dependencies better, can learn multiscale representation of sequential data, and sidestep the gradients problem by using parametrized gating units. This paper proposes a fully connected dense RNN architecture with gated connections to every layer and preceding layer connections, and it's results on PTB charcter-level modelling task."
116,Semantically Decomposing the Latent Spaces of Generative Adversarial Networks,"We propose a new algorithm for training generative adversarial networks to jointly learn latent codes for both identities and observations.In practice, this means that by fixing the identity portion of latent codes, we can generate diverse images of the same subject, and by fixing the observation portion we can traverse the manifold of subjects while maintaining contingent aspects such as lighting and pose.Our algorithm features a pairwise training scheme in which each sample from the generator consists of two images with a common identity code.Corresponding samples from the real dataset consist of two distinct photographs of the same subject.In order to fool the discriminator, the generator must produce images that are both photorealistic, distinct, and appear to depict the same person.We augment both the DCGAN and BEGAN approaches with Siamese discriminators to accommodate pairwise training.Experiments with human judges and an off-the-shelf face verification system demonstrate our algorithm’s ability to generate convincing, identity-matched photographs.","SD-GANs disentangle latent codes according to known commonalities in a dataset (e.g. photographs depicting the same person).This paper investigates the problem of controlled image generation and proposes an algorithm that produces a pair of images with the same identity.This paper proposes, SD-GAN, a method of training GANs to disentangle the identity and non-identity information in the latent vector input Z."
117,AlignFlow: Cycle Consistent Learning from Multiple Domains via Normalizing Flows,"The goal of unpaired cross-domain translation is to learn useful mappings between two domains, given unpaired sets of datapoints from these domains.While this formulation is highly underconstrained, recent work has shown that it is possible to learn mappings useful for downstream tasks by encouraging approximate cycle consistency in the mappings between the two domains [Zhu et al., 2017].In this work, we propose AlignFlow, a framework for unpaired cross-domain translation that ensures exact cycle consistency in the learned mappings.Our framework uses a normalizing flow model to specify a single invertible mapping between the two domains.In contrast to prior works in cycle-consistent translations, we can learn AlignFlow via adversarial training, maximum likelihood estimation, or a hybrid of the two methods.Theoretically, we derive consistency results for AlignFlow which guarantee recovery of desirable mappings under suitable assumptions.Empirically, AlignFlow demonstrates significant improvements over relevant baselines on image-to-image translation and unsupervised domain adaptation tasks on benchmark datasets.","We propose a learning framework for cross-domain translations which is exactly cycle-consistent and can be learned via adversarial training, maximum likelihood estimation, or a hybrid.Proposes AlignFlow, an efficient way of implementing cycle consistency principle using invertible flows.Flow models for unpaired image to image translation"
118,Learning to select examples for program synthesis,"Program synthesis is a class of regression problems where one seeks a solution, in the form of a source-code program, that maps the inputs to their corresponding outputs exactly.Due to its precise and combinatorial nature, it is commonly formulated as a constraint satisfaction problem, where input-output examples are expressed constraints, and solved with a constraint solver.A key challenge of this formulation is that of scalability: While constraint solvers work well with few well-chosen examples, constraining the entire set of example constitutes a significant overhead in both time and memory.In this paper we address this challenge by constructing a representative subset of examples that is both small and is able to constrain the solver sufficiently.We build the subset one example at a time, using a trained discriminator to predict the probability of unchosen input-output examples conditioned on the chosen input-output examples, adding the least probable example to the subset.Experiment on a diagram drawing domain shows our approach produces subset of examples that are small and representative for the constraint solver.","In a program synthesis context where the input is a set of examples, we reduce the cost by computing a subset of representative examplesProposes a method for identifying representative examples for program synthesis to increase the scalability of existing constraint programming solutions.A method for choosing a subset of examples on which to run a constraint solver in order to solve program synthesis problems.This paper proposes a method for speeding up the general-purpose program synthesizers."
119,Recurrent Relational Networks for complex relational reasoning,"Humans possess an ability to abstractly reason about objects and their interactions, an ability not shared with state-of-the-art deep learning models.Relational networks, introduced by Santoro et al., add the capacity for relational reasoning to deep neural networks, but are limited in the complexity of the reasoning tasks they can address.We introduce recurrent relational networks which increase the suite of solvable tasks to those that require an order of magnitude more steps of relational reasoning.We use recurrent relational networks to solve Sudoku puzzles and achieve state-of-the-art results by solving 96.6% of the hardest Sudoku puzzles, where relational networks fail to solve any.We also apply our model to the BaBi textual QA dataset solving 19/20 tasks which is competitive with state-of-the-art sparse differentiable neural computers.The recurrent relational network is a general purpose module that can augment any neural network model with the capacity to do many-step relational reasoning.","We introduce Recurrent Relational Networks, a powerful and general neural network module for relational reasoning, and use it to solve 96.6% of the hardest Sudokus and 19/20 BaBi tasks.Introduced recurrent relational network (RRNs) that can be added to any neural networks to add relational reasoning capacity.Introduction of a deep neural network for structured prediction that achieves state-of-the-art performance on Soduku puzzles and the BaBi task.This paper describes a method called relational network to add relational reasoning capacity to deep neural networks."
120,On the Minimal Supervision for Training Any Binary Classifier from Only Unlabeled Data,"Empirical risk minimization, with proper loss function and regularization, is the common practice of supervised classification.In this paper, we study training arbitrary binary classifier from only unlabeled data by ERM.We prove that it is impossible to estimate the risk of an arbitrary binary classifier in an unbiased manner given a single set of U data, but it becomes possible given two sets of U data with different class priors.These two facts answer a fundamental question what the minimal supervision is for training any binary classifier from only U data.Following these findings, we propose an ERM-based learning method from two sets of U data, and then prove it is consistent.Experiments demonstrate the proposed method could train deep models and outperform state-of-the-art methods for learning from two sets of U data.","Three class priors are all you need to train deep models from only U data, while any two should not be enough.Proposes an unbiased estimator that allows for training models with weak supervision on two unlabeled datasets with known class priors and discusses theoretical properties of the estimators.A methodology for training any binary classifier from only unlabeled data, and an empirical risk minimization method for two sets of unlabeled data where class priors are given."
121,Approximating CNNs with Bag-of-local-Features models works surprisingly well on ImageNet,"Deep Neural Networks excel on many complex perceptual tasks but it has proven notoriously difficult to understand how they reach their decisions.We here introduce a high-performance DNN architecture on ImageNet whose decisions are considerably easier to explain.Our model, a simple variant of the ResNet-50 architecture called BagNet, classifies an image based on the occurrences of small local image features without taking into account their spatial ordering.This strategy is closely related to the bag-of-feature models popular before the onset of deep learning and reaches a surprisingly high accuracy on ImageNet.The constraint on local features makes it straight-forward to analyse how exactly each part of the image influences the classification.Furthermore, the BagNets behave similar to state-of-the art deep neural networks such as VGG-16, ResNet-152 or DenseNet-169 in terms of feature sensitivity, error distribution and interactions between image parts.This suggests that the improvements of DNNs over previous bag-of-feature classifiers in the last few years is mostly achieved by better fine-tuning rather than by qualitatively different decision strategies.","Aggregating class evidence from many small image patches suffices to solve ImageNet, yields more interpretable models and can explain aspects of the decision-making of popular DNNs.This paper suggests a novel and compact neural network architecture which uses the information within bag-of-words features. The proposed algorithm only uses the patch information independently and performs majority voting using independently classified patches."
122,Deep learning mutation prediction enables early stage lung cancer detection in liquid biopsy,"Somatic cancer mutation detection at ultra-low variant allele frequencies is an unmet challenge that is intractable with current state-of-the-art mutation calling methods.Specifically, the limit of VAF detection is closely related to the depth of coverage, due to the requirement of multiple supporting reads in extant methods, precluding the detection of mutations at VAFs that are orders of magnitude lower than the depth of coverage.Nevertheless, the ability to detect cancer-associated mutations in ultra low VAFs is a fundamental requirement for low-tumor burden cancer diagnostics applications such as early detection, monitoring, and therapy nomination using liquid biopsy methods.Here we defined a spatial representation of sequencing information adapted for convolutional architecture that enables variant detection at VAFs, in a manner independent of the depth of sequencing.This method enables the detection of cancer mutations even in VAFs as low as 10x-4^, >2 orders of magnitude below the current state-of-the-art.We validated our method on both simulated plasma and on clinical cfDNA plasma samples from cancer patients and non-cancer controls.This method introduces a new domain within bioinformatics and personalized medicine – somatic whole genome mutation calling for liquid biopsy."," Current somatic mutation methods do not work with liquid biopsies (ie low coverage sequencing), we apply a CNN architecture to a unique representation of a read and its ailgnment, we show significant improvement over previous methods in the low frequency setting.Proposes a CNN based solution called Kittyhawk for somatic mutation calling at ultra low allele frequencies.A new algorithm to detect cancer mutations from sequencing cell free DNA that will identify the sequence context that characterize sequencing errors from true mutations.This paper proposes a deep learning framework to predict somatic mutations at extremely low frequencies which occurs in detecting tumor from cell-free DNA"
123,MedMentions: A Large Biomedical Corpus Annotated with UMLS Concepts,"This paper presents the formal release of, a new manually annotated resource for the recognition of biomedical concepts.What distinguishes MedMentions from other annotated biomedical corpora is its size, as well as the size of the concept ontology and its broad coverage of biomedical disciplines.In addition to the full corpus, a sub-corpus of MedMentions is also presented, comprising annotations for a subset of UMLS 2017 targeted towards document retrieval.To encourage research in Biomedical Named Entity Recognition and Linking, data splits for training and testing are included in the release, and a baseline model and its metrics for entity linking are also described.","The paper introduces a new gold-standard corpus corpus of biomedical scientific literature manually annotated with UMLS concept mentions.Details the construction of a manually annotated dataset covering biomedical concepts that is larger and covered by a larger ontology than previous datasets.This paper uses MedMentions, a TaggerOne semi-Markov model for end-to-end concept recognition and linking on a set of Pubmed abstracts to label papers with biomedical concepts/entities"
124,Sobolev GAN,"We propose a new Integral Probability Metric between distributions: the Sobolev IPM.The Sobolev IPM compares the mean discrepancy of two distributions for functions restricted to a Sobolev ball defined with respect to a dominant measure mu.We show that the Sobolev IPM compares two distributions in high dimensions based on weighted conditional Cumulative Distribution Functions of each coordinate on a leave one out basis.The Dominant measure mu plays a crucial role as it defines the support on which conditional CDFs are compared.Sobolev IPM can be seen as an extension of the one dimensional Von-Mises Cramer statistics to high dimensional distributions.We show how Sobolev IPM can be used to train Generative Adversarial Networks.We then exploit the intrinsic conditioning implied by Sobolev IPM in text generation.Finally we show that a variant of Sobolev GAN achieves competitive results in semi-supervised learning on CIFAR-10, thanks to the smoothness enforced on the critic by Sobolev GAN which relates to Laplacian regularization.","We define a new Integral Probability Metric (Sobolev IPM) and show how it can be used for training GANs for text generation and semi-supervised learning.Suggests a novel regularization scheme for GANs based on a Sobolev norm, measuring deviations between L2 norms of derivatives.The authors provide another type of GAN using the typical setup of a GAN but with a different function class, and produce a recipe for training GANs with that sort of function class.The paper proposes a different gradient penalty for GAN critics that forces the expected squared norm of the gradient to be equal to 1"
125,MGAN: Training Generative Adversarial Nets with Multiple Generators,"We propose in this paper a new approach to train the Generative Adversarial Nets with a mixture of generators to overcome the mode collapsing problem.The main intuition is to employ multiple generators, instead of using a single one as in the original GAN.The idea is simple, yet proven to be extremely effective at covering diverse data modes, easily overcoming the mode collapsing problem and delivering state-of-the-art results.A minimax formulation was able to establish among a classifier, a discriminator, and a set of generators in a similar spirit with GAN.Generators create samples that are intended to come from the same distribution as the training data, whilst the discriminator determines whether samples are true data or generated by generators, and the classifier specifies which generator a sample comes from.The distinguishing feature is that internal samples are created from multiple generators, and then one of them will be randomly selected as final output similar to the mechanism of a probabilistic mixture model.We term our method Mixture Generative Adversarial Nets.We develop theoretical analysis to prove that, at the equilibrium, the Jensen-Shannon divergence between the mixture of generators’ distributions and the empirical data distribution is minimal, whilst the JSD among generators’ distributions is maximal, hence effectively avoiding the mode collapsing problem.By utilizing parameter sharing, our proposed model adds minimal computational cost to the standard GAN, and thus can also efficiently scale to large-scale datasets.We conduct extensive experiments on synthetic 2D data and natural image databases to demonstrate the superior performance of our MGAN in achieving state-of-the-art Inception scores over latest baselines, generating diverse and appealing recognizable objects at different resolutions, and specializing in capturing different types of objects by the generators.","We propose a new approach to train GANs with a mixture of generators to overcome the mode collapsing problem.Address the problem of mode collapse in GANs using a constrained mixture distribution for the generator and an auxiliary classifier which predicts the source mixture component.The paper proposes a mixture of generators to train GANs without extra computational costThe authors present that using MGAN, which aims to overcome model collapsing problem by mixture generators, achieves state-of-art results"
126,BabyAI: A Platform to Study the Sample Efficiency of Grounded Language Learning,"Allowing humans to interactively train artificial agents to understand language instructions is desirable for both practical and scientific reasons. Though, given the lack of sample efficiency in current learning methods, reaching this goal may require substantial research efforts.We introduce the BabyAI research platform, with the goal of supporting investigations towards including humans in the loop for grounded language learning.The BabyAI platform comprises an extensible suite of 19 levels of increasing difficulty.Each level gradually leads the agent towards acquiring a combinatorially rich synthetic language, which is a proper subset of English.The platform also provides a hand-crafted bot agent, which simulates a human teacher. We report estimated amount of supervision required for training neural reinforcement and behavioral-cloning agents on some BabyAI levels.We put forward strong evidence that current deep learning methods are not yet sufficiently sample-efficient in the context of learning a language with compositional properties.",We present the BabyAI platform for studying data efficiency of language learning with a human in the loopPresents a research platform with a bot in the loop for learning to execute language instructions in which language has compositional structuresIntroduces a platform for grounded language learning that replaces any human in the loop with a heuristic teacher and uses a synthetic language mapped to a 2D grid world
127,Deep Learning 3D Shapes Using Alt-az Anisotropic 2-Sphere Convolution,"The ground-breaking performance obtained by deep convolutional neural networks for image processing tasks is inspiring research efforts attempting to extend it for 3D geometric tasks.One of the main challenge in applying CNNs to 3D shape analysis is how to define a natural convolution operator on non-euclidean surfaces.In this paper, we present a method for applying deep learning to 3D surfaces using their spherical descriptors and alt-az anisotropic convolution on 2-sphere.A cascade set of geodesic disk filters rotate on the 2-sphere and collect spherical patterns and so to extract geometric features for various 3D shape analysis tasks.We demonstrate theoretically and experimentally that our proposed method has the possibility to bridge the gap between 2D images and 3D shapes with the desired rotation equivariance/invariance, and its effectiveness is evaluated in applications of non-rigid/ rigid shape classification and shape retrieval.",A method for applying deep learning to 3D surfaces using their spherical descriptors and alt-az anisotropic convolution on 2-sphere.Presents a polar anisotropic convolution scheme on a unit sphere by replacing filter translation with filter rotation.This paper explores deep learning of 3D shapes using alt-az anisotropic 2-sphere convolution
128,Learning Discrete Weights Using the Local Reparameterization Trick,"Recent breakthroughs in computer vision make use of large deep neural networks, utilizing the substantial speedup offered by GPUs.For applications running on limited hardware, however, high precision real-time processing can still be a challenge. One approach to solving this problem is training networks with binary or ternary weights, thus removing the need to calculate multiplications and significantly reducing memory size.In this work, we introduce LR-nets, a new method for training neural networks with discrete weights using stochastic parameters.We show how a simple modification to the local reparameterization trick, previously used to train Gaussian distributed weights, enables the training of discrete weights.Using the proposed training we test both binary and ternary models on MNIST, CIFAR-10 and ImageNet benchmarks and reach state-of-the-art results on most experiments.","Training binary/ternary networks using local reparameterization with the CLT approximationTrains binary and ternary weight distribution networks using backpropagation to sample neuron pre-activations with reparameterization trickThis paper suggests using stochastic parameters in combination with the local reparametrisation trick to train neural networks with binary or ternary weights, which leads to state of the art results."
129,Optimal Completion Distillation for Sequence Learning,"We present Optimal Completion Distillation, a training procedure for optimizing sequence to sequence models based on edit distance.OCD is efficient, has no hyper-parameters of its own, and does not require pre-training or joint optimization with conditional log-likelihood.Given a partial sequence generated by the model, we first identify the set of optimal suffixes that minimize the total edit distance, using an efficient dynamic programming algorithm. Then, for each position of the generated sequence, we use a target distribution which puts equal probability on the first token of all the optimal suffixes.OCD achieves the state-of-the-art performance on end-to-end speech recognition, on both Wall Street Journal and Librispeech datasets, achieving WER and WER, respectively.",Optimal Completion Distillation (OCD) is a training procedure for optimizing sequence to sequence models based on edit distance which achieves state-of-the-art on end-to-end Speech Recognition tasks.Alternative approach to training seq2seq models using a dynamic program to compute optimal continuations of predicted prefixesA training algorithm for auto-regressive models that does not require any MLE pre-training and can directly optimize from the sampling.The paper considers a shortcoming of sequence to sequence models trained using maximum likelihood estimation and propose an approach based on edit distances and the implicit use of given label sequences during training
130,Efficient Federated Learning via Variational Dropout,"As an emerging field, federated learning has recently attracted considerable attention.Compared to distributed learning in the datacenter setting, federated learninghas more strict constraints on computate efficiency of the learned model and communicationcost during the training process.In this work, we propose an efficientfederated learning framework based on variational dropout.Our approach is ableto jointly learn a sparse model while reducing the amount of gradients exchangedduring the iterative training process.We demonstrate the superior performanceof our approach on achieving significant model compression and communicationreduction ratios with no accuracy loss.","a joint model and gradient sparsification method for federated learningApplies variational dropout to reduce the communication cost of distributed training of neural networks, and does experiments on mnist, cifar10 and svhn datasets. The authors propose an algorithm that reduces communication costs in federated learning by sending sparse gradients from device to server and back.Combines distributed optimization algorithm with variational dropout to sparsify the gradients sent to master server from local learners."
131,Learning Deep ResNet Blocks Sequentially using Boosting Theory,"We prove a multiclass boosting theory for the ResNet architectures which simultaneously creates a new technique for multiclass boosting and provides a new algorithm for ResNet-style architectures. Our proposed training algorithm, BoostResNet, is particularly suitable in non-differentiable architectures. Our method only requires the relatively inexpensive sequential training of T ""shallow ResNets"".We prove that the training error decays exponentially with the depth T if the weak module classifiers that we train perform slightly better than some weak baseline. In other words, we propose a weak learning condition and prove a boosting theory for ResNet under the weak learning condition. A generalization error bound based on margin theory is proved and suggests that ResNet could be resistant to overfitting using a network with l_1 norm bounded weights.","We prove a multiclass boosting theory for the ResNet architectures which simultaneously creates a new technique for multiclass boosting and provides a new algorithm for ResNet-style architectures.Presents a boosting-style algorithm for training deep residual networks, a convergence analysis for training error, and a analysis of generalization ability.A learning method for ResNet using the boosting framework that decomposes the learning of complex networks and uses less computational costs.Authors propose the deep ResNet as a boosting algorithm, and they claim this is more efficient than standard end-to-end backropagation."
132,Learning One-hidden-layer Neural Networks with Landscape Design,"We consider the problem of learning a one-hidden-layer neural network: we assume the input x is from Gaussian distribution and the label, where a is a nonnegative vector and is a full-rank weight matrix, and is a noise vector.We first give an analytic formula for the population risk of the standard squared loss and demonstrate that it implicitly attempts to decompose a sequence of low-rank tensors simultaneously.Inspired by the formula, we design a non-convex objective function whose landscape is guaranteed to have the following properties:1.All local minima of are also global minima.2.All global minima of correspond to the ground truth parameters.3.The value and gradient of can be estimated using samples.With these properties, stochastic gradient descent on provably converges to the global minimum and learn the ground-truth parameters.We also prove finite sample complexity results and validate the results by simulations.","The paper analyzes the optimization landscape of one-hidden-layer neural nets and designs a new objective that provably has no spurious local minimum. This paper studies the problem of learning one-hidden layer neural networks, establishes a connection between least squares population loss and Hermite polynomials, and proposes a new loss function.A tensor factorization-type method for leaning one hidden-layer neural netowrk"
133,OPIEC: An Open Information Extraction Corpus,"Open information extraction systems extract relations and their arguments from natural language text in an unsupervised manner.The resulting extractions are a valuable resource for downstream tasks such as knowledge base construction, open question answering, or event schema induction.In this paper, we release, describe, and analyze an OIE corpus called OPIEC, which was extracted from the text of English Wikipedia.OPIEC complements the available OIE resources: It is the largest OIE corpus publicly available to date and contains valuable metadata such as provenance information, confidence scores, linguistic annotations, and semantic annotations including spatial and temporal information.We analyze the OPIEC corpus by comparing its content with knowledge bases such as DBpedia or YAGO, which are also based on Wikipedia.We found that most of the facts between entities present in OPIEC cannot be found in DBpedia and/or YAGO, that OIE facts often differ in the level of specificity compared to knowledge base facts, and that OIE open relations are generally highly polysemous.We believe that the OPIEC corpus is a valuable resource for future research on automated knowledge base construction.",An Open Information Extraction Corpus and its in-depth analysisBuilds a new corpus for information extraction which is larger than the prior public corpora and contains information not existing in current corpora.Presents a dataset of open-IE triples that were collected from Wikipedia with the help of a recent extraction system. The paper describes the creation of an Open IE corpus over English Wikipedia through an automatic manner
134,A Flexible Approach to Automated RNN Architecture Generation,"The process of designing neural architectures requires expert knowledge and extensive trial and error.While automated architecture search may simplify these requirements, the recurrent neural network architectures generated by existing methods are limited in both flexibility and components.We propose a domain-specific language for use in automated architecture search which can produce novel RNNs of arbitrary depth and width.The DSL is flexible enough to define standard architectures such as the Gated Recurrent Unit and Long Short Term Memory and allows the introduction of non-standard RNN components such as trigonometric curves and layer normalization. Using two different candidate generation techniques, random search with a ranking function and reinforcement learning,we explore the novel architectures produced by the RNN DSL for language modeling and machine translation domains.The resulting architectures do not follow human intuition yet perform well on their targeted tasks, suggesting the space of usable RNN architectures is far larger than previously assumed.","We define a flexible DSL for RNN architecture generation that allows RNNs of varying size and complexity and propose a ranking function that represents RNNs as recursive neural networks, simulating their performance to decide on the most promising architectures.Introduces a new method to generate RNNs architectures using a domain-specific language for two types of generators (random and RL-based) together with a ranking function and evaluator.This paper casts the search of good RNN Cell architectures as a black-box optimization problem where examples are represented as an operator tree and scored based on learnt functions or generated by a RL agent.This paper investigates meta-learning strategy for automated architecture search in the context of RNN by using a DSL that specifies RNN recurrent operations."
135,Training and Inference with Integers in Deep Neural Networks,"Researches on deep neural networks with discrete parameters and their deployment in embedded systems have been active and promising topics.Although previous works have successfully reduced precision in inference, transferring both training and inference processes to low-bitwidth integers has not been demonstrated simultaneously.In this work, we develop a new method termed as ""WAGE"" to discretize both training and inference, where weights, activations, gradients and errors among layers are shifted and linearly constrained to low-bitwidth integers.To perform pure discrete dataflow for fixed-point devices, we further replace batch normalization by a constant scaling layer and simplify other components that are arduous for integer implementation.Improved accuracies can be obtained on multiple datasets, which indicates that WAGE somehow acts as a type of regularization.Empirically, we demonstrate the potential to deploy training in hardware systems such as integer-based deep learning accelerators and neuromorphic chips with comparable accuracy and higher energy efficiency, which is crucial to future AI applications in variable scenarios with transfer and continual learning demands.","We apply training and inference with only low-bitwidth integers in DNNsA method called WAGE which quantizes all operands and operators in a neural network to reduce the number of bits for representation in a network.The authors propose discretized weights, activations, gradients, and errors at both training and testing time on neural networks"
136,Fast On-the-fly Retraining-free Sparsification of Convolutional Neural Networks,"Modern Convolutional Neural Networks are complex, encompassing millions of parameters.Their deployment exerts computational, storage and energy demands, particularly on embedded platforms.Existing approaches to prune or sparsify CNNs require retraining to maintain inference accuracy.Such retraining is not feasible in some contexts.In this paper, we explore the sparsification of CNNs by proposing three model-independent methods.Our methods are applied on-the-fly and require no retraining.We show that the state-of-the-art models' weights can be reduced by up to 73% without incurring more than 5% loss in Top-5 accuracy.Additional fine-tuning gains only 8% in sparsity, which indicates that our fast on-the-fly methods are effective.","In this paper, we develop fast retraining-free sparsification methods that can be deployed for on-the-fly sparsification of CNNs in many industrial contexts.This paper proposes approaches for pruning CNNs without retraining by introducing three schemes to determine the thresholds of pruning weights.This paper describes a method for sparsification of CNNs without retraining."
137,Training with Growing Sets: A Simple Alternative to Curriculum Learning and Self Paced Learning,"Curriculum learning and Self paced learning are popular topics in the machine learning that suggest to put the training samples in order by considering their difficulty levels.Studies in these topics show that starting with a small training set and adding new samples according to difficulty levels improves the learning performance.In this paper we experimented that we can also obtain good results by adding the samples randomly without a meaningful order.We compared our method with classical training, Curriculum learning, Self paced learning and their reverse ordered versions.Results of the statistical tests show that the proposed method is better than classical method and similar with the others.These results point a new training regime that removes the process of difficulty level determination in Curriculum and Self paced learning and as successful as these methods.","We propose that training with growing sets stage-by-stage provides an optimization for neural networks.The authors compare curriculum learning to learning in a random order with stages that add a new sample of examples to the previously, randomly constructed setThis paper studies the influence of ordering in the Curriculum and Self paced learning, and shows that to some extent the ordering of training instances is not important."
138,Emerging Disentanglement in Auto-Encoder Based Unsupervised Image Content Transfer,"We study the problem of learning to map, in an unsupervised way, between domains and, such that the samples contain all the information that exists in samples and some additional information.For example, ignoring occlusions, can be people with glasses, people without, and the glasses, would be the added information.When mapping a sample from the first domain to the other domain, the missing information is replicated from an independent reference sample.Thus, in the above example, we can create, for every person without glasses a version with the glasses observed in any face image.Our solution employs a single two-pathway encoder and a single decoder for both domains.The common part of the two domains and the separate part are encoded as two vectors, and the separate part is fixed at zero for domain.The loss terms are minimal and involve reconstruction losses for the two domains and a domain confusion term.Our analysis shows that under mild assumptions, this architecture, which is much simpler than the literature guided-translation methods, is enough to ensure disentanglement between the two domains.We present convincing results in a few visual domains, such as no-glasses to glasses, adding facial hair based on a reference image, etc.","An image to image translation method which adds to one image the content of another thereby creating a new image.This paper tackles the task of content transfer, with the novalty being on the loss."
139,Analysing Mathematical Reasoning Abilities of Neural Models,"Mathematical reasoning a core ability within human intelligence presents some unique challenges as a domain: we do not come to understand and solve mathematical problems primarily on the back of experience and evidence, but on the basis of inferring, learning, and exploiting laws, axioms, and symbol manipulation rules.In this paper, we present a new challenge for the evaluation of neural architectures and similar system, developing a task suite of mathematics problems involving sequential questions and answers in a free-form textual input/output format.The structured nature of the mathematics domain, covering arithmetic, algebra, probability and calculus, enables the construction of training and test spits designed to clearly illuminate the capabilities and failure-modes of different architectures, as well as evaluate their ability to compose and relate knowledge and learned processes.Having described the data generation process and its potential future expansions, we conduct a comprehensive analysis of models from two broad classes of the most powerful sequence-to-sequence architectures and find notable differences in their ability to resolve mathematical problems and generalize their knowledge.","A dataset for testing mathematical reasoning (and algebraic generalization), and results on current sequence-to-sequence models.Presents a new synthetic dataset to evaluate the mathematical reasoning ability of sequence-to-sequence models, and uses it to evaluate several models.Model for solving basic math problems."
140,Low-Cost Parameterizations of Deep Convolutional Neural Networks,"Convolutional Neural Networks filter the input data using a series of spatial convolution operators with compactly supported stencils and point-wise nonlinearities.Commonly, the convolution operators couple features from all channels.For wide networks, this leads to immense computational cost in the training of and prediction with CNNs.In this paper, we present novel ways to parameterize the convolution more efficiently, aiming to decrease the number of parameters in CNNs and their computational complexity.We propose new architectures that use a sparser coupling between the channels and thereby reduce both the number of trainable weights and the computational cost of the CNN.Our architectures arise as new types of residual neural network that can be seen as discretizations of a Partial Differential Equations and thus have predictable theoretical properties.Our first architecture involves a convolution operator with a special sparsity structure, and is applicable to a large class of CNNs.Next, we present an architecture that can be seen as a discretization of a diffusion reaction PDE, and use it with three different convolution operators.We outline in our experiments that the proposed architectures, although considerably reducing the number of trainable weights, yield comparable accuracy to existing CNNs that are fully coupled in the channel dimension.","This paper introduces efficient and economic parametrizations of convolutional neural networks motivated by partial differential equations Introduces four ""low cost"" alternatives to the standard convolution operation that can be used in place of the standard convolution operation to reduce their computational complexity.This paper introduces methods for reducing the computational cost of CNN implementations, and introduces new parameterizations of CNN like architectures that limit parameter coupling.The paper proposes a PDE-based perspective to understand and parameterize CNNs"
141,Information Theoretic lower bounds on negative log likelihood,"In this article we use rate-distortion theory, a branch of information theory devoted to the problem of lossy compression, to shed light on an important problem in latent variable modeling of data: is there room to improve the model?One way to address this question is to find an upper bound on the probability that the model can assign to some data as one varies the prior and/or the likelihood function in a latent variable model.The core of our contribution is to formally show that the problem of optimizing priors in latent variable models is exactly an instance of the variational optimization problem that information theorists solve when computing rate-distortion functions, and then to use this to derive a lower bound on negative log likelihood.Moreover, we will show that if changing the prior can improve the log likelihood, then there is a way to change the likelihood function instead and attain the same log likelihood, and thus rate-distortion theory is of relevance to both optimizing priors as well as optimizing likelihood functions.We will experimentally argue for the usefulness of quantities derived from rate-distortion theory in latent variable modeling by applying them to a problem in image modeling.",Use rate-distortion theory to bound how much a latent variable model can be improvedAddresses problems of optimization of the prior in the latent variable model and the selection of the likelihood function by proposing criteria based on a lower-bound on the negative log-likelihood.Presents a theorem which gives a lower bound on negative log likelihood of rate-distortion for latent-variable modelingThe authors argue that the rate-distortion theory for lossy compression provides a natural toolkit for studying latent variable models proposes a lower bound.
142,Linear Backprop in non-linear networks,"Backprop is the primary learning algorithm used in many machine learning algorithms.In practice, however, Backprop in deep neural networks is a highly sensitive learning algorithm and successful learning depends on numerous conditions and constraints.One set of constraints is to avoid weights that lead to saturated units.The motivation for avoiding unit saturation is that gradients vanish and as a result learning comes to a halt.Careful weight initialization and re-scaling schemes such as batch normalization ensure that input activity to the neuron is within the linear regime where gradients are not vanished and can flow.Here we investigate backpropagating error terms only linearly.That is, we ignore the saturation that arise by ensuring gradients always flow.We refer to this learning rule as Linear Backprop since in the backward pass the network appears to be linear.In addition to ensuring persistent gradient flow, Linear Backprop is also favorable when computation is expensive since gradients are never computed.Our early results suggest that learning with Linear Backprop is competitive with Backprop and saves expensive gradient computations.",We ignore non-linearities and do not compute gradients in the backward pass to save computation and to ensure gradients always flow. The author proposed linear backprop algorithms to ensure gradients flow for all parts during backpropagation.
143,Towards a better understanding of Vector Quantized Autoencoders,"Deep neural networks with discrete latent variables offer the promise of better symbolic reasoning, and learning abstractions that are more useful to new tasks.There has been a surge in interest in discrete latent variable models, however, despite several recent improvements, the training of discrete latent variable models has remained challenging and their performance has mostly failed to match their continuous counterparts.Recent work on vector quantized autoencoders has made substantial progress in this direction, with its perplexity almost matching that of a VAE on datasets such as CIFAR-10.In this work, we investigate an alternate training technique for VQ-VAE, inspired by its connection to the Expectation Maximization algorithm.Training the discrete autoencoder with EM and combining it with sequence level knowledge distillation alows us to develop a non-autoregressive machine translation model whose accuracy almost matches a strong greedy autoregressive baseline Transformer, while being 3.3 times faster at inference.",Understand the VQ-VAE discrete autoencoder systematically using EM and use it to design non-autogressive translation model matching a strong autoregressive baseline.This paper introduces a new way of interpreting the VQ-VAE and proposes a new training algorithm based on the soft EM clustering.The paper presents an alternative view on the training procedure for the VQ-VAE using the soft EM algorithm
144,Optimal margin Distribution Network,"Recent research about margin theory has proved that maximizing the minimum margin like support vector machines does not necessarily lead to better performance, and instead, it is crucial to optimize the margin distribution.In the meantime, margin theory has been used to explain the empirical success of deep network in recent studies.In this paper, we present ODN, a network which embeds a loss function in regard to the optimal margin distribution.We give a theoretical analysis for our method using the PAC-Bayesian framework, which confirms the significance of the margin distribution for classification within the framework of deep networks.In addition, empirical results show that the ODN model always outperforms the baseline cross-entropy loss model consistently across different regularization situations.And our ODNmodel also outperforms the cross-entropy loss, hinge loss and soft hinge loss model in generalization task through limited training data.","This paper presents a deep neural network embedding a loss function in regard to the optimal margin distribution, which alleviates the overfitting problem theoretically and empirically.Presents a PAC-Bayesian bound for a margin loss"
145,Deep Net Triage: Assessing The Criticality of Network Layers by Structural Compression,"Deep network compression seeks to reduce the number of parameters in the network while maintaining a certain level of performance. Deep network distillation seeks to train a smaller network that matches soft-max performance of a larger network. While both regimes have led to impressive performance for their respective goals, neither provide insight into the importance of a given layer in the original model, which is useful if we are to improve our understanding of these highly parameterized models. In this paper, we present the concept of deep net triage, which individually assesses small blocks of convolution layers to understand their collective contribution to the overall performance, which we call . We call it triage because we assess this criticality by answering the question: what is the impact to the health of the overall network if we compress a block of layers into a single layer.We propose a suite of triage methods and compare them on problem spaces of varying complexity. We ultimately show that, across these problem spaces, deep net triage is able to indicate the of relative importance of different layers. Surprisingly, our local structural compression technique also leads to an improvement in overall accuracy when the final model is fine-tuned globally.",We seek to understand learned representations in compressed networks via an experimental regime we call deep net triageCompares various initialization and training methods of transferring knowledge from VGG network to a smaller student network by replacing blocks of layers with single layers.This paper presents five methods for doing triaging or block layer compression for deep networks.The paper proposes a method to compress a block of layers in a NN that evaluates several different sub-approaches
146,Super-Convergence: Very Fast Training of Residual Networks Using Large Learning Rates,"In this paper, we show a phenomenon, which we named super-convergence'', where residual networks can be trained using an order of magnitude fewer iterations than is used with standard training methods. "", 'The existence of super-convergence is relevant to understanding why deep networks generalize well. One of the key elements of super-convergence is training with cyclical learning rates and a large maximum learning rate. Furthermore, we present evidence that training with large learning rates improves performance by regularizing the network.In addition, we show that super-convergence provides a greater boost in performance relative to standard training when the amount of labeled training data is limited. We also derive a simplification of the Hessian Free optimization method to compute an estimate of the optimal learning rate. The architectures to replicate this work will be made available upon publication.","Empirical proof of a new phenomenon requires new theoretical insights and is relevent to the active discussions in the literature on SGD and understanding generalization.The paper discusses a phenomenon where neural network training in very specific settings can profit much from a schedule including large learning ratesThe authors analyze training of residual networks using large cyclic learning rates, and demonstrate fast convergence with cyclic learning rates and evidence of large learning rates acting as regularization."
147,Infinitely Deep Infinite-Width Networks,"Infinite-width neural networks have been extensively used to study the theoretical properties underlying the extraordinary empirical success of standard, finite-width neural networks.Nevertheless, until now, infinite-width networks have been limited to at most two hidden layers.To address this shortcoming, we study the initialisation requirements of these networks and show that the main challenge for constructing them is defining the appropriate sampling distributions for the weights.Based on these observations, we propose a principled approach to weight initialisation that correctly accounts for the functional nature of the hidden layer activations and facilitates the construction of arbitrarily many infinite-width layers, thus enabling the construction of arbitrarily deep infinite-width networks.The main idea of our approach is to iteratively reparametrise the hidden-layer activations into appropriately defined reproducing kernel Hilbert spaces and use the canonical way of constructing probability distributions over these spaces for specifying the required weight distributions in a principled way.Furthermore, we examine the practical implications of this construction for standard, finite-width networks.In particular, we derive a novel weight initialisation scheme for standard, finite-width networks that takes into account the structure of the data and information about the task at hand.We demonstrate the effectiveness of this weight initialisation approach on the MNIST, CIFAR-10 and Year Prediction MSD datasets.","We propose a method for the construction of arbitrarily deep infinite-width networks, based on which we derive a novel weight initialisation scheme for finite-width networks and demonstrate its competitive performance.Proposes a weight initialization approach to enable infinitely deep and infinite-width networks with experimental results on small datasets.Proposes deep neural networks of infinite width."
148,A Self-Organizing Memory Network,"Working memory requires information about external stimuli to be represented in the brain even after those stimuli go away.This information is encoded in the activities of neurons, and neural activities change over timescales of tens of milliseconds.Information in working memory, however, is retained for tens of seconds, suggesting the question of how time-varying neural activities maintain stable representations.Prior work shows that, if the neural dynamics are in the ` null space' of the representation - so that changes to neural activity do not affect the downstream read-out of stimulus information - then information can be retained for periods much longer than the time-scale of individual-neuronal activities.The prior work, however, requires precisely constructed synaptic connectivity matrices, without explaining how this would arise in a biological neural network.To identify mechanisms through which biological networks can self-organize to learn memory function, we derived biologically plausible synaptic plasticity rules that dynamically modify the connectivity matrix to enable information storing.Networks implementing this plasticity rule can successfully learn to form memory representations even if only 10% of the synapses are plastic, they are robust to synaptic noise, and they can represent information about multiple stimuli.","We derived biologically plausible synaptic plasticity learning rules for a recurrent neural network to store stimulus representations. A neural network model consisting of recurrently connected neurons and one or more redouts which aims to retain some output over time.This paper presents a self-organizing memory mechanism in a neural model, and introduces an objective function that minimizes changes in the signal to be memorized."
149,On the limitations of first order approximation in GAN dynamics,"Generative Adversarial Networks have been proposed as an approach to learning generative models.While GANs have demonstrated promising performance on multiple vision tasks, their learning dynamics are not yet well understood, neither in theory nor in practice.In particular, the work in this domain has been focused so far only on understanding the properties of the stationary solutions that this dynamics might converge to, and of the behavior of that dynamics in this solutions’ immediate neighborhood.To address this issue, in this work we take a first step towards a principled study of the GAN dynamics itself.To this end, we propose a model that, on one hand, exhibits several of the common problematic convergence behaviors, but on the other hand, is sufficiently simple to enable rigorous convergence analysis.This methodology enables us to exhibit an interesting phenomena: a GAN with an optimal discriminator provably converges, while guiding the GAN training using only a first order approximation of the discriminator leads to unstable GAN dynamics and mode collapse.This suggests that such usage of the first order approximation of the discriminator, which is a de-facto standard in all the existing GAN dynamics, might be one of the factors that makes GAN training so challenging in practice.Additionally, our convergence result constitutes the first rigorous analysis of a dynamics of a concrete parametric GAN.","To understand GAN training, we define simple GAN dynamics, and show quantitative differences between optimal and first order updates in this model.The authors study the impact of GANs in settings where at each iteration, the discriminator trains to convergence and the generator updates with gradient steps, or where a few gradient steps are done for the disciminator and generator.This paper studies the dynamics of adversarial training of GANs on a Gaussian mixture model"
150,GradMix: Multi-source Transfer across Domains and Tasks,"The machine learning and computer vision community is witnessing an unprecedented rate of new tasks being proposed and addressed, thanks to the power of deep convolutional networks to find complex mappings from X to Y. The advent of each task often accompanies the release of a large-scale human-labeled dataset, for supervised training of the deep network.However, it is expensive and time-consuming to manually label sufficient amount of training data.Therefore, it is important to develop algorithms that can leverage off-the-shelf labeled dataset to learn useful knowledge for the target task.While previous works mostly focus on transfer learning from a single source, we study multi-source transfer across domains and tasks, in a semi-supervised setting.We propose GradMix, a model-agnostic method applicable to any model trained with gradient-based learning rule.GradMix transfers knowledge via gradient descent, by weighting and mixing the gradients from all sources during training.Our method follows a meta-learning objective, by assigning layer-wise weights to the source gradients, such that the combined gradient follows the direction that can minimize the loss for a small set of samples from the target dataset.In addition, we propose to adaptively adjust the learning rate for each mini-batch based on its importance to the target task, and a pseudo-labeling method to leverage the unlabeled samples in the target domain.We perform experiments on two MS-DTT tasks: digit recognition and action recognition, and demonstrate the advantageous performance of the proposed method against multiple baselines.",We propose a gradient-based method to transfer knowledge from multiple sources across different domains and tasks.This paper proposes to combine the gradients of source domains to help the learning in the target domain. 
151,Variational Bayesian Phylogenetic Inference,"Bayesian phylogenetic inference is currently done via Markov chain Monte Carlo with simple mechanisms for proposing new states, which hinders exploration efficiency and often requires long runs to deliver accurate posterior estimates.In this paper we present an alternative approach: a variational framework for Bayesian phylogenetic analysis.We approximate the true posterior using an expressive graphical model for tree distributions, called a subsplit Bayesian network, together with appropriate branch length distributions.We train the variational approximation via stochastic gradient ascent and adopt multi-sample based gradient estimators for different latent variables separately to handle the composite latent space of phylogenetic models.We show that our structured variational approximations are flexible enough to provide comparable posterior estimation to MCMC, while requiring less computation due to a more efficient tree exploration mechanism enabled by variational inference.Moreover, the variational approximations can be readily used for further statistical analysis such as marginal likelihood estimation for model comparison via importance sampling.Experiments on both synthetic data and real data Bayesian phylogenetic inference problems demonstrate the effectiveness and efficiency of our methods.","The first variational Bayes formulation of phylogenetic inference, a challenging inference problem over structures with intertwined discrete and continuous componentsExplores an approximate inference solution to the problem of Bayesian inference of phylogenetic trees by leveraging recently proposed subsplit Bayesian networks and modern gradient estimators for VI.Proposes a variational approach to Bayesian posterior inference in phylogenetic trees."
152,HybridNet: A Hybrid Neural Architecture to Speed-up Autoregressive  Models,"This paper introduces HybridNet, a hybrid neural network to speed-up autoregressivemodels for raw audio waveform generation.As an example, we proposea hybrid model that combines an autoregressive network named WaveNet and aconventional LSTM model to address speech synthesis.Instead of generatingone sample per time-step, the proposed HybridNet generates multiple samples pertime-step by exploiting the long-term memory utilization property of LSTMs.Inthe evaluation, when applied to text-to-speech, HybridNet yields state-of-art performance.HybridNet achieves a 3.83 subjective 5-scale mean opinion score onUS English, largely outperforming the same size WaveNet in terms of naturalnessand provide 2x speed up at inference.","It is a hybrid neural architecture to speed-up autoregressive model. Concludes that in order to scale up the model size without increasing inference time for sequential prediction, use a model that predicts multiple timesteps at once.This paper presents HybridNet, a neural speech and other audio synthesis system that combines the WaveNet model with an LSTM with the goal of offering a model with faster inference-time audio generation."
153,Visual Explanation by Interpretation: Improving Visual Feedback Capabilities of Deep Neural Networks,"Visual Interpretation and explanation of deep models is critical towards wide adoption of systems that rely on them.In this paper, we propose a novel scheme for both interpretation as well as explanation in which, given a pretrained model, we automatically identify internal features relevant for the set of classes considered by the model, without relying on additional annotations.We interpret the model through average visualizations of this reduced set of features.Then, at test time, we explain the network prediction by accompanying the predicted class label with supporting visualizations derived from the identified features.In addition, we propose a method to address the artifacts introduced by strided operations in deconvNet-based visualizations.Moreover, we introduce an8Flower , a dataset specifically designed for objective quantitative evaluation of methods for visual explanation.Experiments on the MNIST , ILSVRC 12, Fashion 144k and an8Flower datasets show that our method produces detailed explanations with good coverage of relevant features of the classes of interest.","Interpretation by Identifying model-learned features that serve as indicators for the task of interest. Explain model decisions by highlighting the response of these features in test data. Evaluate explanations objectively with a controlled dataset.This paper proposes a method for producing visual explanations for deep neural network outputs and releases a new synthetic dataset.A method for Deep Neural Networks that identifies automatically relevant features of the set of the classes, supporting interpretation and explanation without relying on additional annotations."
154,An efficient framework for learning sentence representations,"In this work we propose a simple and efficient framework for learning sentence representations from unlabelled data.Drawing inspiration from the distributional hypothesis and recent work on learning sentence representations, we reformulate the problem of predicting the context in which a sentence appears as a classification problem.Given a sentence and the context in which it appears, a classifier distinguishes context sentences from other contrastive sentences based on their vector representations.This allows us to efficiently learn different types of encoding functions, and we show that the model learns high-quality sentence representations.We demonstrate that our sentence representations outperform state-of-the-art unsupervised and supervised representation learning methods on several downstream NLP tasks that involve understanding sentence semantics while achieving an order of magnitude speedup in training time.",A framework for learning high-quality sentence representations efficiently.Proposes a faster algorithm for learning SkipThought-style sentence representations from corpora of ordered sentences that swaps the word-level decoder for a contrastive classification loss.This paper proposes a framework for unsupervised learning of sentence representations by maximizing a model of the probability of true context sentences relative to random candidate sentences
155,From Information Bottleneck To Activation Norm Penalty,"Many regularization methods have been proposed to prevent overfitting in neural networks.Recently, a regularization method has been proposed to optimize the variational lower bound of the Information Bottleneck Lagrangian.However, this method cannot be generalized to regular neural network architectures.We present the activation norm penalty that is derived from the information bottleneck principle and is theoretically grounded in a variation dropout framework.Unlike in previous literature, it can be applied to any general neural network.We demonstrate that this penalty can give consistent improvements to different state of the art architectures both in language modeling and image classification.We present analyses on the properties of this penalty and compare it to other methods that also reduce mutual information.","We derive a norm penalty on the output of the neural network from the information bottleneck perspectivePuts forward Activation Norm Penalty, an L_2 type regularization on the activations, deriving it from the Information Bottleneck principleThis paper creates a mapping between activation norm penalties and information bottleneck framework using variational dropout framework."
156,Deep Temporal Clustering: Fully unsupervised learning of time-domain features,"Unsupervised learning of timeseries data is a challenging problem in machine learning.Here,we propose a novel algorithm, Deep Temporal Clustering, a fully unsupervised method, to naturally integrate dimensionality reduction and temporal clustering into a single end to end learning framework.The algorithm starts with an initial cluster estimates using an autoencoder for dimensionality reduction and a novel temporal clustering layer for cluster assignment.Then it jointly optimizes the clustering objective and the dimensionality reduction objective.Based on requirement and application, the temporal clustering layer can be customized with any temporal similarity metric.Several similarity metrics are considered and compared. To gain insight into features that the network has learned for its clustering, we apply a visualization method that generates a heat map of regions of interest in the timeseries.The viability of the algorithm is demonstrated using timeseries data from diverse domains, ranging from earthquakes to sensor data from spacecraft.In each case, we show that our algorithm outperforms traditional methods.This performance is attributed to fully integrated temporal dimensionality reduction and clustering criterion.","A fully unsupervised method, to naturally integrate dimensionality reduction and temporal clustering into a single end to end learning framework.Proposes an algorithm that integrates autoencoder with time-series data clustering using a network structure that suits time-series data.An algorithm for jointly performing dimensionality reduction and temporal clustering in a deep learning context, utilizing an autoencoder and clustering objective.The authors proposed an unsupervised time series clustering methods built with deep neural networks and equipped with an encoder-decoder and a clustering mode to shorten the time series, extract local temporal features, and to get the encoded representations."
157,MahiNet: A Neural Network for Many-Class Few-Shot Learning with Class Hierarchy,"We study many-class few-shot problem in both supervised learning and meta-learning scenarios.Compared to the well-studied many-class many-shot and few-class few-shot problems, MCFS problem commonly occurs in practical applications but is rarely studied.MCFS brings new challenges because it needs to distinguish between many classes, but only a few samples per class are available for training.In this paper, we propose memory-augmented hierarchical-classification network'' for MCFS learning."", ""It addresses the many-class'' problem by exploring the class hierarchy, e.g., the coarse-class label that covers a subset of fine classes, which helps to narrow down the candidates for the fine class and is cheaper to obtain.MahiNet uses a convolutional neural network to extract features, and integrates a memory-augmented attention module with a multi-layer perceptron to produce the probabilities over coarse and fine classes.While the MLP extends the linear classifier, the attention module extends a KNN classifier, both together targeting the ''`few-shot'' problem.We design different training strategies of MahiNet for supervised learning and meta-learning.Moreover, we propose two novel benchmark datasets ''mcfsImageNet'' and ''mcfsOmniglot'' specifically for MCFS problem.In experiments, we show that MahiNet outperforms several state-of-the-art models on MCFS classification tasks in both supervised learning and meta-learning scenarios.",A memory-augmented neural network that addresses many-class few-shot problem by leveraging class hierarchy in both supervised learning and meta-learning.This paper presents methods for adding inductive bias to a classifier through coarse-to-fine prediction along a class hierarchy and learning a memory-based KNN classifier that keeps track of mislabeled instances during learning.This paper formulates the many-class-few-shot classification problem from a supervised learning perspective and a meta-learning perspective.
158,Forced Apart: Discovering Disentangled Representations Without Exhaustive Labels,"Learning a better representation with neural networks is a challenging problem, which has been tackled from different perspectives in the past few years.In this work, we focus on learning a representation that would be useful in a clustering task.We introduce two novel loss components that substantially improve the quality of produced clusters, are simple to apply to arbitrary models and cost functions, and do not require a complicated training procedure.We perform an extensive set of experiments, supervised and unsupervised, and evaluate the proposed loss components on two most common types of models, Recurrent Neural Networks and Convolutional Neural Networks, showing that the approach we propose consistently improves the quality of KMeans clustering in terms of mutual information scores and outperforms previously proposed methods.",A novel loss component that forces the network to learn a representation that is well-suited for clustering during training for a classification task.This paper proposes two regularization terms based on a compound hinge loss over the KL divergence between two softmax-normalized input arguments to encourage learning disentangled representationsProposal for two regularizers intended to make the representations learned in the penultimate layer of a classifier more conforming to inherent structure in the data.
159,Learning Representations for Faster Similarity Search,"In high dimensions, the performance of nearest neighbor algorithms depends crucially on structure in the data.While traditional nearest neighbor datasets consisted mostly of hand-crafted feature vectors, an increasing number of datasets comes from representations learned with neural networks.We study the interaction between nearest neighbor algorithms and neural networks in more detail.We find that the network architecture can significantly influence the efficacy of nearest neighbor algorithms even when the classification accuracy is unchanged.Based on our experiments, we propose a number of training modifications that lead to significantly better datasets for nearest neighbor algorithms.Our modifications lead to learned representations that can accelerate nearest neighbor queries by 5x.",We show how to get good representations from the point of view of Simiarity Search.Studies the impact of changing the image classification part on top of the DNN on the ability to index the descriptors with a LSH or a kd-tree algorithm.Proposes to use softmax cross-entropy loss to learn a network that tries to reduce the angles between inputs and the corresponding class vectors in a supervised framework using.
160,Relaxed Quantization for Discretized Neural Networks,"Neural network quantization has become an important research area due to its great impact on deployment of large models on resource constrained devices.In order to train networks that can be effectively discretized without loss of performance, we introduce a differentiable quantization procedure.Differentiability can be achieved by transforming continuous distributions over the weights and activations of the network to categorical distributions over the quantization grid.These are subsequently relaxed to continuous surrogates that can allow for efficient gradient-based optimization.We further show that stochastic rounding can be seen as a special case of the proposed approach and that under this formulation the quantization grid itself can also be optimized with gradient descent.We experimentally validate the performance of our method on MNIST, CIFAR 10 and Imagenet classification.",We introduce a technique that allows for gradient based training of quantized neural networks.Proposes a unified and general way of training neural networks with reduced precision quantized synaptic weights and activations.A new approach to quantizing activations which is state of the art or competitive on several real image problems.A method for learning neural networks with quantized weights and activations by stochastically quantizing values and replacing the resulting categotical distribution with a continuous relaxation
161,Distributional Adversarial Networks,"In most current formulations of adversarial training, the discriminators can be expressed as single-input operators, that is, the mapping they define is separable over observations.In this work, we argue that this property might help explain the infamous mode collapse phenomenon in adversarially-trained generative models.Inspired by discrepancy measures and two-sample tests between probability distributions, we propose distributional adversaries that operate on samples, i.e., on sets of multiple points drawn from a distribution, rather than on single observations.We show how they can be easily implemented on top of existing models.Various experimental results show that generators trained in combination with our distributional adversaries are much more stable and are remarkably less prone to mode collapse than traditional models trained with observation-wise prediction discriminators.In addition, the application of our framework to domain adaptation results in strong improvement over recent state-of-the-art.","We show that the mode collapse problem in GANs may be explained by a lack of information sharing between observations in a training batch, and propose a distribution-based framework for globally sharing information between gradients that leads to more stable and effective adversarial training.Proposes to replace single-sample discriminators in adversarial training with discriminators that explicitly operate on distributions of examples.Theory on two-sample tests and MMD and how can be beneficially incorporated into GAN framework."
162,CHEMICAL NAMES STANDARDIZATION USING NEURAL SEQUENCE TO SEQUENCE MODEL,"Chemical information extraction is to convert chemical knowledge in text into true chemical database, which is a text processing task heavily relying on chemical compound name identification and standardization.Once a systematic name for a chemical compound is given, it will naturally and much simply convert the name into the eventually required molecular formula.However, for many chemical substances, they have been shown in many other names besides their systematic names which poses a great challenge for this task.In this paper, we propose a framework to do the auto standardization from the non-systematic names to the corresponding systematic names by using the spelling error correction, byte pair encoding tokenization and neural sequence to sequence model.Our framework is trained end to end and is fully data-driven.Our standardization accuracy on the test dataset achieves 54.04% which has a great improvement compared to previous state-of-the-art result.",We designed an end-to-end framework using sequence to sequence model to do the chemical names standardization.Standardizes non systematic names in chemical information extraction by creating a parallel corpus of non-systematic and systematic names and building a seq2seq model.This work presents a method to translate non-systematic names of chemical compounds into their systematic equivalents using a combination of mechanisms
163,On the Relation Between the Sharpest Directions of DNN Loss and the SGD Step Length,"The training of deep neural networks with Stochastic Gradient Descent with a large learning rate or a small batch-size typically ends in flat regions of the weight space, as indicated by small eigenvalues of the Hessian of the training loss.This was found to correlate with a good final generalization performance. In this paper we extend previous work by investigating the curvature of the loss surface along the whole training trajectory, rather than only at the endpoint.We find that initially SGD visits increasingly sharp regions, reaching a maximum sharpness determined by both the learning rate and the batch-size of SGD.At this peak value SGD starts to fail to minimize the loss along directions in the loss surface corresponding to the largest curvature.To further investigate the effect of these dynamics in the training process, we study a variant of SGD using a reduced learning rate along the sharpest directions which we show can improve training speed while finding both sharper and better generalizing solution, compared to vanilla SGD.Overall, our results show that the SGD dynamics in the subspace of the sharpest directions influence the regions that SGD steers to, the overall training speed, and the generalization ability of the final model.","SGD is steered early on in training towards a region in which its step is too large compared to curvature, which impacts the rest of training. Analyzes the relationship between the convergence/generalization and the update on largest eigenvectors of Hessian of the empirical losses of DNNs.This work studies the relationship between the SGD step size and the curvature of the loss surface"
164,Predicting Multiple Actions for Stochastic Continuous Control,"We introduce a new approach to estimate continuous actions using actor-critic algorithms for reinforcement learning problems.Policy gradient methods usually predict one continuous action estimate or parameters of a presumed distribution for any given state which might not be optimal as it may not capture the complete description of the target distribution.Our approach instead predicts M actions with the policy network and then uniformly sample one action during training as well as testing at each state.This allows the agent to learn a simple stochastic policy that has an easy to compute expected return.In all experiments, this facilitates better exploration of the state space during training and converges to a better policy.","We introduce a novel reinforcement learning algorithm, that predicts multiple actions and samples from them.This work introduces a uniform mixture of deterministic policies, and find that this parametrization of stochastic policies outperforms DDPG on several OpenAI gym benchmarks.The authors investigate a method for improving the performance of networks trained with DDPG, and show improved performance on a large number of standard continuous control environment."
165,Reconciling Feature-Reuse and Overfitting in DenseNet with Specialized Dropout,"Recently convolutional neural networks achieve great accuracy in visual recognition tasks.DenseNet becomes one of the most popular CNN models due to its effectiveness in feature-reuse.However, like other CNN models, DenseNets also face overfitting problem if not severer.Existing dropout method can be applied but not as effective due to the introduced nonlinear connections.In particular, the property of feature-reuse in DenseNet will be impeded, and the dropout effect will be weakened by the spatial correlation inside feature maps.To address these problems, we craft the design of a specialized dropout method from three aspects, dropout location, dropout granularity, and dropout probability.The insights attained here could potentially be applied as a general approach for boosting the accuracy of other CNN models with similar nonlinear connections.Experimental results show that DenseNets with our specialized dropout method yield better accuracy compared to vanilla DenseNet and state-of-the-art CNN models, and such accuracy boost increases with the model depth.","Realizing the drawbacks when applying original dropout on DenseNet, we craft the design of dropout method from three aspects, the idea of which could also be applied on other CNN models.Application of different binary dropout structures and schedules with the specific aim to regularise the DenseNet architecture.Proposes a pre-dropout technique for densenet which implements the dropout before the non-linear activation function."
166,Human-Guided Column Networks: Augmenting Deep Learning with Advice,"While extremely successful in several applications, especially with low-level representations; sparse, noisy samples and structured domains are some of the open challenges in most deep models.Column Networks, a deep architecture, can succinctly capture such domain structure and interactions, but may still be prone to sub-optimal learning from sparse and noisy samples.Inspired by the success of human-advice guided learning in AI, especially in data-scarce domains, we propose Knowledge-augmented Column Networks that leverage human advice/knowledge for better learning with noisy/sparse samples.Our experiments demonstrate how our approach leads to either superior overall performance or faster convergence.","Guiding relation-aware deep models towards better learning with human knowledge.This work proposes a variant of the column network based on the injection of human guidance by modifying calculations in the network.A method to incorporate human advices to deep learning by extending Column Network, a graph neural network for collective classification."
167,The High-Dimensional Geometry of Binary Neural Networks,"Recent research has shown that one can train a neural network with binary weights and activations at train time by augmenting the weights with a high-precision continuous latent variable that accumulates small changes from stochastic gradient descent.However, there is a dearth of work to explain why one can effectively capture the features in data with binary weights and activations.Our main result is that the neural networks with binary weights and activations trained using the method of Courbariaux, Hubara et al. work because of the high-dimensional geometry of binary vectors.In particular, the ideal continuous vectors that extract out features in the intermediate representations of these BNNs are well-approximated by binary vectors in the sense that dot products are approximately preserved.Compared to previous research that demonstrated good classification performance with BNNs, our work explains why these BNNs work in terms of HD geometry. Furthermore, the results and analysis used on BNNs are shown to generalize to neural networks with ternary weights and activations.Our theory serves as a foundation for understanding not only BNNs but a variety of methods that seek to compress traditional neural networks.Furthermore, a better understanding of multilayer binary neural networks serves as a starting point for generalizing BNNs to other neural network architectures such as recurrent neural networks.",Recent successes of Binary Neural Networks can be understood based on the geometry of high-dimensional binary vectorsInvestigates numerically and theoretically the reasons behind the empirical success of binarized neural networks.This paper analyzes the effectiveness of binary neural networks and why binarization is able to preserve model performance.
168,CNNs as Inverse Problem Solvers and Double Network Superresolution,"In recent years Convolutional Neural Networks have been used extensively for Superresolution.In this paper, we use inverse problem and sparse representation solutions to form a mathematical basis for CNN operations.We show how a single neuron is able to provide the optimum solution for inverse problem, given a low resolution image dictionary as an operator.Introducing a new concept called Representation Dictionary Duality, we show that CNN elements are trained to be representation vectors and then, during reconstruction, used as dictionaries.In the light of theoretical work, we propose a new algorithm which uses two networks with different structures that are separately trained with low and high coherency image patches and show that it performs faster compared to the state-of-the-art algorithms while not sacrificing from performance.","After proving that a neuron acts as an inverse problem solver for superresolution and a network of neurons is guarantied to provide a solution, we proposed a double network architecture that performs faster than state-of-the-art.Discusses using neural networks for super-resolutionA new architecture for solving image super-resolution tasks, and an analysis aiming to establish a connection between CNNs for solving super resolution and solving sparse regularized inverse problems."
169,Divide and Conquer Networks,"We consider the learning of algorithmic tasks by mere observation of input-outputpairs.Rather than studying this as a black-box discrete regression problem withno assumption whatsoever on the input-output mapping, we concentrate on tasksthat are amenable to the principle of divide and conquer, and study what are itsimplications in terms of learning.This principle creates a powerful inductive bias that we leverage with neuralarchitectures that are defined recursively and dynamically, by learning two scale-invariant atomic operations: how to split a given input into smaller sets, and howto merge two partially solved tasks into a larger partial solution.Our model can betrained in weakly supervised environments, namely by just observing input-outputpairs, and in even weaker environments, using a non-differentiable reward signal.Moreover, thanks to the dynamic aspect of our architecture, we can incorporatethe computational complexity as a regularization term that can be optimized bybackpropagation.We demonstrate the flexibility and efficiency of the Divide-and-Conquer Network on several combinatorial and geometric tasks: convex hull,clustering, knapsack and euclidean TSP.Thanks to the dynamic programmingnature of our model, we show significant improvements in terms of generalizationerror and computational complexity.","Dynamic model that learns divide and conquer strategies by weak supervision.Proposes to add new inductive bias to neural network architecture by using a divide and conquer strategy.This paper studies problems that can be solved using a dynamic programming approach, and proposes a neural network architecture to solve such problems that beats sequence to sequence baselines.The paper proposes a unique network architecture that can learn divide-and-conquer strategies to solve algorithmic tasks."
170,GO Gradient for Expectation-Based Objectives,"Within many machine learning algorithms, a fundamental problem concerns efficient calculation of an unbiased gradient wrt parameters for expectation-based objectives.Most existing methods either suffer from high variance, seeking help from complicated variance-reduction techniques; or they only apply to reparameterizable continuous random variables and employ a reparameterization trick.To address these limitations, we propose a General and One-sample gradient that applies to many distributions associated with non-reparameterizable continuous discrete random variables, and has the same low-variance as the reparameterization trick.We find that the GO gradient often works well in practice based on only one Monte Carlo sample.Alongside the GO gradient, we develop a means of propagating the chain rule through distributions, yielding statistical back-propagation, coupling neural networks to common random variables.","a Rep-like gradient for non-reparameterizable continuous/discrete distributions; further generalized to deep probabilistic models, yielding statistical back-propagationPresents a gradient estimator for expectation-based objectives that is unbiased, has low variance, and applies to either continuous and discrete random variables.An improved method for computing derivates of the expectation, and a new gradient estimator of low variance that allows training of generative models in which observations or latent variables are discrete.Designs a low variance gradient for distributions associated with continuous or discrete random variables."
171,Cost-Sensitive Robustness against Adversarial Examples,"Several recent works have developed methods for training classifiers that are certifiably robust against norm-bounded adversarial perturbations.These methods assume that all the adversarial transformations are equally important, which is seldom the case in real-world applications.We advocate for cost-sensitive robustness as the criteria for measuring the classifier's performance for tasks where some adversarial transformation are more important than others.We encode the potential harm of each adversarial transformation in a cost matrix, and propose a general objective function to adapt the robust training method of Wong & Kolter to optimize for cost-sensitive robustness.Our experiments on simple MNIST and CIFAR10 models with a variety of cost matrices show that the proposed approach can produce models with substantially reduced cost-sensitive robust error, while maintaining classification accuracy.",A general method for training certified cost-sensitive robust classifier against adversarial perturbationsCalculates and plugs in the costs of adversarial attack into the objective of optimization to get a model that is cost-sensitively robust against adversarial attacks. Build on semnial work by Dalvi et al. and extends approach to certifiable robustness with a cost matrix that specifies for each pair of source-target classes whether the model should be robust to adversarial examples.
172,Learning a neural response metric for retinal prosthesis,"Retinal prostheses for treating incurable blindness are designed to electrically stimulate surviving retinal neurons, causing them to send artificial visual signals to the brain.However, electrical stimulation generally cannot precisely reproduce normal patterns of neural activity in the retina.Therefore, an electrical stimulus must be selected that produces a neural response as close as possible to the desired response.This requires a technique for computing a distance between the desired response and the achievable response that is meaningful in terms of the visual signal being conveyed.Here we propose a method to learn such a metric on neural responses, directly from recorded light responses of a population of retinal ganglion cells in the primate retina.The learned metric produces a measure of similarity of RGC population responses that accurately reflects the similarity of the visual input.Using data from electrical stimulation experiments, we demonstrate that this metric may improve the performance of a prosthesis.","Using triplets to learn a metric for comparing neural responses and improve the performance of a prosthesis.Authors develop new spike train distance metrics, including neural networks and quadratic metrics. These metrics are shown to outperform the naive Hamming distance metric, and implicitly captures some structure in neural code.With the application of improving neural prosthesis in mind, the authors propose to learn a metric between neural responses by either optimizing a quadratic form or a deep neural network ."
173,QCue: Queries and Cues for Computer-Facilitated Mind-Mapping,"We introduce a novel workflow, QCue, for providing textual stimulation during mind-mapping.Mind-mapping is a powerful tool whose intent is to allow one to externalize ideas and their relationships surrounding a central problem.The key challenge in mind-mapping is the difficulty in balancing the exploration of different aspects of the problem with a detailed exploration of each of those aspects.Our idea behind QCue is based on two mechanisms: computer-generated automatic cues to stimulate the user to explore the breadth of topics based on the temporal and topological evolution of a mind-map and user-elicited queries for helping the user explore the depth for a given topic.We present a two-phase study wherein the first phase provided insights that led to the development of our work-flow for stimulating the user through cues and queries.In the second phase, we present a between-subjects evaluation comparing QCue with a digital mind-mapping work-flow without computer intervention.Finally, we present an expert rater evaluation of the mind-maps created by users in conjunction with user feedback.","This paper introduces a method to generate questions (cues) and queries (suggestions) to help users perform mind-mapping.Presents a tool to assist mind-mapping through suggested context related to existing nodes and through questions that expand on less developed branches.This paper presents an approach for assisting people with mindmapping tasks, designing an interface and algorithmic features to suppport mindmapping, and contributes a evaluative study."
174,Detecting Out-Of-Distribution Samples Using Low-Order Deep Features Statistics,"The ability to detect when an input sample was not drawn from the training distribution is an important desirable property of deep neural networks.In this paper, we show that a simple ensembling of first and second order deep feature statistics can be exploited to effectively differentiate in-distribution and out-of-distribution samples.Specifically, we observe that the mean and standard deviation within feature maps differs greatly between in-distribution and out-of-distribution samples.Based on this observation, we propose a simple and efficient plug-and-play detection procedure that does not require re-training, pre-processing or changes to the model. The proposed method outperforms the state-of-the-art by a large margin in all standard benchmarking tasks, while being much simpler to implement and execute.Notably, our method improves the true negative rate from 39.6% to 95.3% when 95% of in-distribution are correctly detected using a DenseNet and the out-of-distribution dataset is TinyImageNet resize.The source code of our method will be made publicly available.",Detecting out-of-distribution samples by using low-order feature statistics without requiring any change in underlying DNN.Presents an algorithm to detect out-of-distribution samples by using the running estimate of mean and variance within BatchNorm layers to construct feature representations later fed into a linear classifier.An approach for detecting out-of-distribution samples in which the authors propose to use logistic regression over simple statistics of each batch normalization layer of CNN.The paper suggests using Z-scores for comparing ID and OOD samples to evaluate what deep nets are trying to do.
175,Maximal Divergence Sequential Autoencoder for Binary Software Vulnerability Detection,"Due to the sharp increase in the severity of the threat imposed by software vulnerabilities, the detection of vulnerabilities in binary code has become an important concern in the software industry, such as the embedded systems industry, and in the field of computer security.However, most of the work in binary code vulnerability detection has relied on handcrafted features which are manually chosen by a select few, knowledgeable domain experts.In this paper, we attempt to alleviate this severe binary vulnerability detection bottleneck by leveraging recent advances in deep learning representations and propose the Maximal Divergence Sequential Auto-Encoder.In particular, latent codes representing vulnerable and non-vulnerable binaries are encouraged to be maximally divergent, while still being able to maintain crucial information from the original binaries.We conducted extensive experiments to compare and contrast our proposed methods with the baselines, and the results show that our proposed methods outperform the baselines in all performance measures of interest.","We propose a novel method named Maximal Divergence Sequential Auto-Encoder that leverages Variational AutoEncoder representation for binary code vulnerability detection.This paper proposes a variational autoencoder-based architecture for code embeddings for binary software vulnerability detection, with learned embeddings more effective at distinguishing between vulnerable and non-vulnerable binary code than baselines.This paper proposes a model to automatically extract features for vulnerability detection using deep learning technique."
176,Posterior Attention Models for Sequence to Sequence Learning,"Modern neural architectures critically rely on attention for mapping structured inputs to sequences.In this paper we show that prevalent attention architectures do not adequately model the dependence among the attention and output tokens across a predicted sequence.We present an alternative architecture called Posterior Attention Models that after a principled factorization of the full joint distribution of the attention and output variables, proposes two major changes. First, the position where attention is marginalized is changed from the input to the output.Second, the attention propagated to the next decoding stage is a posterior attention distribution conditioned on the output.Empirically on five translation and two morphological inflection tasks the proposed posterior attention models yield better BLEU score and alignment accuracy than existing attention models.","Computing attention based on posterior distribution leads to more meaningful attention and better performanceThis paper proposes a sequence to sequence model where attention is treated as a latent variable, and derives novel inference procedures for this model, obtaining improvements in machine translation and morphological inflection generation tasks.This paper presents a novel posterior attention model for seq2seq problems"
177,DNN Model Compression Under Accuracy Constraints,"The growing interest to implement Deep Neural Networks on resource-bound hardware has motivated innovation of compression algorithms.Using these algorithms, DNN model sizes can be substantially reduced, with little to no accuracy degradation.This is achieved by either eliminating components from the model, or penalizing complexity during training.While both approaches demonstrate considerable compressions, the former often ignores the loss function during compression while the later produces unpredictable compressions.In this paper, we propose a technique that directly minimizes both the model complexity and the changes in the loss function.In this technique, we formulate compression as a constrained optimization problem, and then present a solution for it.We will show that using this technique, we can achieve competitive results.",Compressing trained DNN models by minimizing their complexity while constraining their loss.This paper proposes a method for deep neural network compression under accuracy constraints.This paper presents a loss value constrained k-means encoding method for network compression and develops an iterative algorithm for model optimization.
178,Modeling Latent Attention Within Neural Networks,"Deep neural networks are able to solve tasks across a variety of domains and modalities of data.Despite many empirical successes, we lack the ability to clearly understand and interpret the learned mechanisms that contribute to such effective behaviors and more critically, failure modes.In this work, we present a general method for visualizing an arbitrary neural network's inner mechanisms and their power and limitations.Our dataset-centric method produces visualizations of how a trained network attends to components of its inputs.The computed ""attention masks"" support improved interpretability by highlighting which input attributes are critical in determining output.We demonstrate the effectiveness of our framework on a variety of deep neural network architectures in domains from computer vision and natural language processing.The primary contribution of our approach is an interpretable visualization of attention that provides unique insights into the network's underlying decision-making process irrespective of the data modality.",We develop a technique to visualize attention mechanisms in arbitrary neural networks. Proposes to learn a Latent Attention Network that can help to visualize the inner structure of a deep neural network.The authors of this paper propose a data-driven black-box visualization scheme. 
179,Exploring Deep Recurrent Models with Reinforcement Learning for Molecule Design,"The design of small molecules with bespoke properties is of central importance to drug discovery. However significant challenges yet remain for computational methods, despite recent advances such as deep recurrent networks and reinforcement learning strategies for sequence generation, and it can be difficult to compare results across different works. This work proposes 19 benchmarks selected by subject experts, expands smaller datasets previously used to approximately 1.1 million training molecules, and explores how to apply new reinforcement learning techniques effectively for molecular design. The benchmarks here, built as OpenAI Gym environments, will be open-sourced to encourage innovation in molecular design algorithms and to enable usage by those without a background in chemistry. Finally, this work explores recent development in reinforcement-learning methods with excellent sample complexity and investigates their behavior in molecular generation, demonstrating significant performance gains compared to standard reinforcement learning techniques.","We investigate a variety of RL algorithms for molecular generation and define new benchmarks (to be released as an OpenAI Gym), finding PPO and a hill-climbing MLE algorithm work best.Considers model evaluation for molecule generation by proposing 19 benchmarks, expanding small data sets to a large, standardized dataset, and exploring how to apply RL techniques for molecular design.This paper shows that the most sophisticated RL methods are less effective than the simple hill-climbing technique, with PPO as the exception, when modeling and synthesizing molecules."
180,Learning to Make Analogies by Contrasting Abstract Relational Structure,"Analogical reasoning has been a principal focus of various waves of AI research.Analogy is particularly challenging for machines because it requires relational structures to be represented such that they can be flexibly applied across diverse domains of experience.Here, we study how analogical reasoning can be induced in neural networks that learn to perceive and reason about raw visual data.We find that the critical factor for inducing such a capacity is not an elaborate architecture, but rather, careful attention to the choice of data and the manner in which it is presented to the model.The most robust capacity for analogical reasoning is induced when networks learn analogies by contrasting abstract relational structures in their input domains, a training method that uses only the input data to force models to learn about important abstract features.Using this technique we demonstrate capacities for complex, visual and symbolic analogy making and generalisation in even the simplest neural network architectures.","The most robust capacity for analogical reasoning is induced when networks learn analogies by contrasting abstract relational structures in their input domains.The paper investigates the ability of a neural network to learn analogy, showing that a simple neural network is able to solve certain analogy problemsThis paper describes an approach to train neural networks for analogical reasoning tasks, specifically considering visual analogy and symbolic analogies."
181,A Goal-oriented Neural Conversation Model by Self-Play,"Building chatbots that can accomplish goals such as booking a flight ticket is an unsolved problem in natural language understanding.Much progress has been made to build conversation models using techniques such as sequence2sequence modeling.One challenge in applying such techniques to building goal-oriented conversation models is that maximum likelihood-based models are not optimized toward accomplishing goals.Recently, many methods have been proposed to address this issue by optimizing a reward that contains task status or outcome.However, adding the reward optimization on the fly usually provides little guidance for language construction and the conversation model soon becomes decoupled from the language model.In this paper, we propose a new setting in goal-oriented dialogue system to tighten the gap between these two aspects by enforcing model level information isolation on individual models between two agents.Language construction now becomes an important part in reward optimization since it is the only way information can be exchanged.We experimented our models using self-play and results showed that our method not only beat the baseline sequence2sequence model in rewards but can also generate human-readable meaningful conversations of comparable quality.","A Goal-oriented Neural Conversation Model by Self-PlayA self-play model for goal oriented dialog generation, aiming to enforce a stronger coupling between the task reward and the language model.This paper describes a method for improving a goal oriented dialogue system using selfplay. "
182,Realtime query completion via deep language models,"Search engine users nowadays heavily depend on query completion and correction to shape their queries. Typically, the completion is done by database lookup which does not understand the context and cannot generalize to prefixes not in the database. In the paper, we propose to use unsupervised deep language models to complete and correct the queries given an arbitrary prefix. We show how to address two main challenges that renders this method practical for large-scale deployment: 1) we propose a method for integrating error correction into the language model completion via a edit-distance potential and a variant of beam search that can exploit these potential functions; and2) we show how to efficiently perform CPU-based computation to complete the queries, with error correction, in real time.Experiments show that the method substantially increases hit rate over standard approaches, and is capable of handling tail queries.","realtime search query completion using character-level LSTM language modelsThis paper presents methods for query completion that includes prefix correction, and some engineering details to meet particular latency requirements on a CPU.The authors propose an algorithm for solving the query completion problem with error correction, and adopt character-level RNN-based modeling and optimize the inference part to achieve targets in real time."
183,Convergence Guarantees for RMSProp and ADAM in Non-Convex Optimization and an Empirical Comparison to Nesterov Acceleration,"RMSProp and ADAM continue to be extremely popular algorithms for training neural nets but their theoretical convergence properties have remained unclear.Further, recent work has seemed to suggest that these algorithms have worse generalization properties when compared to carefully tuned stochastic gradient descent or its momentum variants.In this work, we make progress towards a deeper understanding of ADAM and RMSProp in two ways.First, we provide proofs that these adaptive gradient algorithms are guaranteed to reach criticality for smooth non-convex objectives, and we give bounds on the running time.Next we design experiments to empirically study the convergence and generalization properties of RMSProp and ADAM against Nesterov's Accelerated Gradient method on a variety of common autoencoder setups and on VGG-9 with CIFAR-10.Through these experiments we demonstrate the interesting sensitivity that ADAM has to its momentum parameter beta_1.We show that at very high values of the momentum parameter ADAM outperforms a carefully tuned NAG on most of our experiments, in terms of getting lower training and test losses.On the other hand, NAG can sometimes do better when ADAM's beta_1 is set to the most commonly used value: beta_1 = 0.9, indicating the importance of tuning the hyperparameters of ADAM to get better generalization performance.We also report experiments on different autoencoders to demonstrate that NAG has better abilities in terms of reducing the gradient norms, and it also produces iterates which exhibit an increasing trend for the minimum eigenvalue of the Hessian of the loss function at the iterates.",In this paper we prove convergence to criticality of (stochastic and deterministic) RMSProp and deterministic ADAM for smooth non-convex objectives and we demonstrate an interesting beta_1 sensitivity for ADAM on autoencoders. This paper presents a convergence analysis of RMSProp and ADAM in the case of smooth non-convex functions
184,Towards Safe Deep Learning: Unsupervised Defense Against Generic Adversarial Attacks,"Recent advances in adversarial Deep Learning have opened up a new and largely unexplored surface for malicious attacks jeopardizing the integrity of autonomous DL systems.We introduce a novel automated countermeasure called Parallel Checkpointing Learners to thwart the potential adversarial attacks and significantly improve the reliability of a victim DL model.The proposed PCL methodology is unsupervised, meaning that no adversarial sample is leveraged to build/train parallel checkpointing learners.We formalize the goal of preventing adversarial attacks as an optimization problem to minimize the rarely observed regions in the latent feature space spanned by a DL network.To solve the aforementioned minimization problem, a set of complementary but disjoint checkpointing modules are trained and leveraged to validate the victim model execution in parallel.Each checkpointing learner explicitly characterizes the geometry of the input data and the corresponding high-level data abstractions within a particular DL layer.As such, the adversary is required to simultaneously deceive all the defender modules in order to succeed.We extensively evaluate the performance of the PCL methodology against the state-of-the-art attack scenarios, including Fast-Gradient-Sign, Jacobian Saliency Map Attack, Deepfool, and Carlini&WagnerL2 algorithm.Extensive proof-of-concept evaluations for analyzing various data collections including MNIST, CIFAR10, and ImageNet corroborate the effectiveness of our proposed defense mechanism against adversarial samples.",Devising unsupervised defense mechanisms against adversarial attacks is crucial to ensure the generalizability of the defense. This paper presents a method for detecting adversarial examples in a deep learning classification settingThis paper presents an unsupervised method for detecting adversarial examples of neural networks.
185,ProxylessNAS: Direct Neural Architecture Search on Target Task and Hardware,"Neural architecture search has a great impact by automatically designing effective neural network architectures.However, the prohibitive computational demand of conventional NAS algorithms makes it difficult to directly search the architectures on large-scale tasks.Differentiable NAS can reduce the cost of GPU hours via a continuous representation of network architecture but suffers from the high GPU memory consumption issue.As a result, they need to utilize proxy tasks, such as training on a smaller dataset, or learning with only a few blocks, or training just for a few epochs.These architectures optimized on proxy tasks are not guaranteed to be optimal on the target task.In this paper, we present ProxylessNAS that can directly learn the architectures for large-scale target tasks and target hardware platforms.We address the high memory consumption issue of differentiable NAS and reduce the computational cost to the same level of regular training while still allowing a large candidate set.Experiments on CIFAR-10 and ImageNet demonstrate the effectiveness of directness and specialization.On CIFAR-10, our model achieves 2.08% test error with only 5.7M parameters, better than the previous state-of-the-art architecture AmoebaNet-B, while using 6× fewer parameters.On ImageNet, our model achieves 3.1% better top-1 accuracy than MobileNetV2, while being 1.2× faster with measured GPU latency.We also apply ProxylessNAS to specialize neural architectures for hardware with direct hardware metrics and provide insights for efficient CNN architecture design.","Proxy-less neural architecture search for directly learning architectures on large-scale target task (ImageNet) while reducing the cost to the same level of normal training.This paper addresses the problem of architecture search, and specifically seeks to do this without having to train on ""proxy"" tasks where the problem is simplified through more limited optimization, architectural complexity, or dataset size."
186,A Variational Dirichlet Framework for Out-of-Distribution Detection,"With the recently rapid development in deep learning, deep neural networks have been widely adopted in many real-life applications.However, deep neural networks are also known to have very little control over its uncertainty for test examples, which potentially causes very harmful and annoying consequences in practical scenarios.In this paper, we are particularly interested in designing a higher-order uncertainty metric for deep neural networks and investigate its performance on the out-of-distribution detection task proposed by~.Our method first assumes there exists a underlying higher-order distribution, which generated label-wise distributionover classes on the K-dimension simplex, and then approximate such higher-order distribution via parameterized posterior function under variational inference framework, finally we use the entropy of learned posterior distribution as uncertainty measure to detect out-of-distribution examples. However, we identify the overwhelming over-concentration issue in such a framework, which greatly hinders the detection performance. Therefore, we further design a log-smoothing function to alleviate such issue to greatly increase the robustness of the proposed entropy-based uncertainty measure. Throughcomprehensive experiments on various datasets and architectures, our proposed variational Dirichlet framework with entropy-based uncertainty measure is consistently observed to yield significant improvements over many baseline systems.","A new framework based variational inference for out-of-distribution detectionDescribes a probabilistic approach to quantifying uncertainty in DNN classification tasks that outperforms other SOTA methods in the task of out-of-distribution detection.A new framework for out-of-distribution detection, based on variaitonal inference and a prior Dirichlet distribution, that reports state of the art results on several datasets.An out-of distribution detection via a new method to approximate the confidence distribution of classification probability using variational inference of Dirichlet distribution."
187,Learning what you can do before doing anything,"Intelligent agents can learn to represent the action spaces of other agents simply by observing them act.Such representations help agents quickly learn to predict the effects of their own actions on the environment and to plan complex action sequences.In this work, we address the problem of learning an agent’s action space purely from visual observation.We use stochastic video prediction to learn a latent variable that captures the scene's dynamics while being minimally sensitive to the scene's static content.We introduce a loss term that encourages the network to capture the composability of visual sequences and show that it leads to representations that disentangle the structure of actions.We call the full model with composable action representations Composable Learned Action Space Predictor.We show the applicability of our method to synthetic settings and its potential to capture action spaces in complex, realistic visual settings.When used in a semi-supervised setting, our learned representations perform comparably to existing fully supervised methods on tasks such as action-conditioned video prediction and planning in the learned action space, while requiring orders of magnitude fewer action labels.Project website: https://daniilidis-group.github.io/learned_action_spaces","We learn a representation of an agent's action space from pure visual observations. We use a recurrent latent variable approach with a novel composability loss."", 'Proposes a compositional latent-variable model to learn models that predict what will happen next in scenarios where action-labels are not available in abundance.A variational IB based approach to learn action representations directly from videos of actions being taken, achieving better efficiency of subsequent learning methods while requiring lesser amount of action label videos.This paper proposes an approach to video prediction which autonomously finds an action space encoding differences between subsequent frames"
188,Negotiating Team Formation Using Deep Reinforcement Learning,"When autonomous agents interact in the same environment, they must often cooperate to achieve their goals.One way for agents to cooperate effectively is to form a team, make a binding agreement on a joint plan, and execute it.However, when agents are self-interested, the gains from team formation must be allocated appropriately to incentivize agreement.Various approaches for multi-agent negotiation have been proposed, but typically only work for particular negotiation protocols.More general methods usually require human input or domain-specific data, and so do not scale.To address this, we propose a framework for training agents to negotiate and form teams using deep reinforcement learning.Importantly, our method makes no assumptions about the specific negotiation protocol, and is instead completely experience driven.We evaluate our approach on both non-spatial and spatially extended team-formation negotiation environments, demonstrating that our agents beat hand-crafted bots and reach negotiation outcomes consistent with fair solutions predicted by cooperative game theory.Additionally, we investigate how the physical location of agents influences negotiation outcomes.","Reinforcement learning can be used to train agents to negotiate team formation across many negotiation protocolsThis paper studies deep multi-agent RL in settings where all of the agents must cooperate to accomplish a task (e.g., search and rescue, multi-player video games), and uses simple cooperative weighted voting games to study the efficacy of deep RL and to compare solutions found by deep RL to a fair solution.A reinforcement learning approach for negotiating coalitions in cooperative game theory settings that can be used in cases where unlimited training simulations are available."
189,Identifying and Controlling Important Neurons in Neural Machine Translation,"Neural machine translation models learn representations containing substantial linguistic information.However, it is not clear if such information is fully distributed or if some of it can be attributed to individual neurons.We develop unsupervised methods for discovering important neurons in NMT models.Our methods rely on the intuition that different models learn similar properties, and do not require any costly external supervision.We show experimentally that translation quality depends on the discovered neurons, and find that many of them capture common linguistic phenomena.Finally, we show how to control NMT translations in predictable ways, by modifying activations of individual neurons.","Unsupervised methods for finding, analyzing, and controlling important neurons in NMTThis paper presents unsupervised approaches to discovering important neurons in neural machine translation systems and analyzes linguistic properties controlled by those neurons.Unsupervised methods for ranking neurons in machine translation where important neurons are thus identified and used to control the MT output."
190,Universal Agent for Disentangling Environments and Tasks,"Recent state-of-the-art reinforcement learning algorithms are trained under the goal of excelling in one specific task.Hence, both environment and task specific knowledge are entangled into one framework.However, there are often scenarios where the environment is fixed while only the target task changes.Hence, borrowing the idea from hierarchical reinforcement learning, we propose a framework that disentangles task and environment specific knowledge by separating them into two units.The environment-specific unit handles how to move from one state to the target state; and the task-specific unit plans for the next target state given a specific task.The extensive results in simulators indicate that our method can efficiently separate and learn two independent units, and also adapt to a new task more efficiently than the state-of-the-art methods.","We propose a DRL framework that disentangles task and environment specific knowledge.The authors propose to decompose reinforcement learning into a PATH function and a GOAL functionA modular architecture with the aim of separating environment specific knowledge and task-specific knowledge into different modules, on par with standard A3C across a wide range of tasks."
191,Pix2Scene: Learning Implicit 3D Representations from Images,"Modelling 3D scenes from 2D images is a long-standing problem in computer vision with implications in, e.g., simulation and robotics.We propose pix2scene, a deep generative-based approach that implicitly models the geometric properties of a scene from images.Our method learns the depth and orientation of scene points visible in images.Our model can then predict the structure of a scene from various, previously unseen view points.It relies on a bi-directional adversarial learning mechanism to generate scene representations from a latent code, inferring the 3D representation of the underlying scene geometry.We showcase a novel differentiable renderer to train the 3D model in an end-to-end fashion, using only images.We demonstrate the generative ability of our model qualitatively on both a custom dataset and on ShapeNet.Finally, we evaluate the effectiveness of the learned 3D scene representation in supporting a 3D spatial reasoning.","pix2scene: a deep generative based approach for implicitly modelling the geometrical properties of a 3D scene from imagesExplores explaining scenes with surfels in a neural recognition model, and demonstrate results on image reconstruction, synthesis, and mental shape rotation.Authors introduce a method to create a 3D scene model given a 2D image and a camera pose using a self-superfised model"
192,Learning Relation Representations from Word Representations,"Identifying the relations that connect words is an important step towards understanding human languages and is useful for various NLP tasks such as knowledge base completion and analogical reasoning.Simple unsupervised operators such as vector offset between two-word embeddings have shown to recover some specific relationships between those words, if any.Despite this, how to accurately learn generic relation representations from word representations remains unclear.We model relation representation as a supervised learning problem and learn parametrised operators that map pre-trained word embeddings to relation representations.We propose a method for learning relation representations using a feed-forward neural network that performs relation prediction.Our evaluations on two benchmark datasets reveal that the penultimate layer of the trained neural network-based relational predictor acts as a good representation for the relations between words.",Identifying the relations that connect words is important for various NLP tasks. We model relation representation as a supervised learning problem and learn parametrised operators that map pre-trained word embeddings to relation representations.This paper presents a novel method for representing lexical relations as vectors using just pre-trained word embeddings and a novel loss function operating over pairs of word pairs.A novel solution to the relation compositon problem when you already have pre trained word/entity embeddings and are interested only in learning to compose relation representations.
193,Fraternal Dropout,"Recurrent neural networks are important class of architectures among neural networks useful for language modeling and sequential prediction.However, optimizing RNNs is known to be harder compared to feed-forward neural networks.A number of techniques have been proposed in literature to address this problem.In this paper we propose a simple technique called fraternal dropout that takes advantage of dropout to achieve this goal.Specifically, we propose to train two identical copies of an RNN with different dropout masks while minimizing the difference between their predictions.In this way our regularization encourages the representations of RNNs to be invariant to dropout mask, thus being robust.We show that our regularization term is upper bounded by the expectation-linear dropout objective which has been shown to address the gap due to the difference between the train and inference phases of dropout.We evaluate our model and achieve state-of-the-art results in sequence modeling tasks on two benchmark datasets - Penn Treebank and Wikitext-2.We also show that our approach leads to performance improvement by a significant margin in image captioning and semi-supervised tasks.","We propose to train two identical copies of an recurrent neural network (that share parameters) with different dropout masks while minimizing the difference between their (pre-softmax) predictions.Presents Fraternal dropout as an improvement over Expectation-linear dropout in terms of convergence, and demonstrates the utility of Fraternal dropout on a number of tasks and datasets."
194,Generating Liquid Simulations with Deformation-aware Neural Networks,"We propose a novel approach for deformation-aware neural networks that learn the weighting and synthesis of dense volumetric deformation fields.Our method specifically targets the space-time representation of physical surfaces from liquid simulations.Liquids exhibit highly complex, non-linear behavior under changing simulation conditions such as different initial conditions.Our algorithm captures these complex phenomena in two stages: a first neural network computes a weighting function for a set of pre-computed deformations, while a second network directly generates a deformation field for refining the surface.Key for successful training runs in this setting is a suitable loss function that encodes the effect of the deformations, and a robust calculation of the corresponding gradients.To demonstrate the effectiveness of our approach, we showcase our method with several complex examples of flowing liquids with topology changes.Our representation makes it possible to rapidly generate the desired implicit surfaces.We have implemented a mobile application to demonstrate that real-time interactions with complex liquid effects are possible with our approach.",Learning weighting and deformations of space-time data sets for highly efficient approximations of liquid behavior.A neural-network based model is used to interpolate simulations for novel scene conditions from densely registered 4D implicit surfaces for a structured scene.This paper presents a coupled deep learning approach for generating realistic liquid simulation data that can be useful for real-time decision support applications.This paper introduces a deep learning approach for physical simulation that combines two networks for synthesizing 4D data that represents 3D physical simulations
195,On the Construction and Evaluation of Color Invariant Networks,"This is an empirical paper which constructs color invariant networks and evaluates their performances on a realistic data set.The paper studies the simplest possible case of color invariance: invariance under pixel-wise permutation of the color channels.Thus the network is aware not of the specific color object, but its colorfulness.The data set introduced in the paper consists of images showing crashed cars from which ten classes were extracted.An additional annotation was done which labeled whether the car shown was red or non-red. The networks were evaluated by their performance on the classification task.With the color annotation we altered the color ratios in the training data and analyzed the generalization capabilities of the networks on the unaltered test data.We further split the test data in red and non-red cars and did a similar evaluation.It is shown in the paper that an pixel-wise ordering of the rgb-values of the images performs better or at least similarly for small deviations from the true color ratios.The limits of these networks are also discussed.","We construct and evaluate color invariant neural nets on a novel realistic data setProposes a method to make neural networks for image recognition color invariant and evaluates it on the cifar 10 dataset.The authors investigate a modified input layer that results in color invariant networks, and show that certain color invariant input layers can improve accuracy for test-images from a different color distribution than the training images.The authors test a CNN on images with color channels modified to be invariant to permutations, with performance not degraded by too much. "
196,On the Expressive Power of Overlapping Architectures of Deep Learning,"Expressive efficiency refers to the relation between two architectures A and B, whereby any function realized by B could be replicated by A, but there exists functions realized by A, which cannot be replicated by B unless its size grows significantly larger.For example, it is known that deep networks are exponentially efficient with respect to shallow networks, in the sense that a shallow network must grow exponentially large in order to approximate the functions represented by a deep network of polynomial size.In this work, we extend the study of expressive efficiency to the attribute of network connectivity and in particular to the effect of ""overlaps"" in the convolutional process, i.e., when the stride of the convolution is smaller than its filter size.To theoretically analyze this aspect of network's design, we focus on a well-established surrogate for ConvNets called Convolutional Arithmetic Circuits, and then demonstrate empirically that our results hold for standard ConvNets as well.Specifically, our analysis shows that having overlapping local receptive fields, and more broadly denser connectivity, results in an exponential increase in the expressive capacity of neural networks.Moreover, while denser connectivity can increase the expressive capacity, we show that the most common types of modern architectures already exhibit exponential increase in expressivity, without relying on fully-connected layers.","We analyze how the degree of overlaps between the receptive fields of a convolutional network affects its expressive power.The paper studies the expressive power provided by ""overlap"" in convolution layers of DNNs by considering linear activations with product pooling.This paper analyzes the expressivity of convolutional arithmetic circuits and shows that an exponentialy large number of non-overlapping ConvACs are required to approximate the grid tensor of an overlapping ConvACs."
197,Efficiently testing local optimality and escaping saddles for ReLU networks,"We provide a theoretical algorithm for checking local optimality and escaping saddles at nondifferentiable points of empirical risks of two-layer ReLU networks.Our algorithm receives any parameter value and returns: local minimum, second-order stationary point, or a strict descent direction.The presence of M data points on the nondifferentiability of the ReLU divides the parameter space into at most 2^M regions, which makes analysis difficult.By exploiting polyhedral geometry, we reduce the total computation down to one convex quadratic program for each hidden node, Oequality tests, and one nonconvex QP.For the last QP, we show that our specific problem can be solved efficiently, in spite of nonconvexity.In the benign case, we solve one equality constrained QP, and we prove that projected gradient descent solves it exponentially fast.In the bad case, we have to solve a few more inequality constrained QPs, but we prove that the time complexity is exponential only in the number of inequality constraints.Our experiments show that either benign case or bad case with very few inequality constraints occurs, implying that our algorithm is efficient in most cases.","A theoretical algorithm for testing local optimality and extracting descent directions at nondifferentiable points of empirical risks of one-hidden-layer ReLU networks.Proposes an algorithm to check whether a given point is a generalized second-order stationary point.A theoretical algorithm, involving solving convex and non-convex quadratic programs, for checking local optimality and escaping saddles when training two-layer ReLU networks.Author proposes a method to check if a point is a stationary point or not and then classify stationary points as either local min or second-order stationary"
198,VSE++: Improving Visual-Semantic Embeddings with Hard Negatives,"We present a new technique for learning visual-semantic embeddings for cross-modal retrieval. Inspired by the use of hard negatives in structured prediction, and ranking loss functions used in retrieval, we introduce a simple change to common loss functions used to learn multi-modal embeddings. That, combined with fine-tuning and the use of augmented data, yields significant gains in retrieval performance. We showcase our approach, dubbed VSE++, on the MS-COCO and Flickr30K datasets, using ablation studies and comparisons with existing methods. On MS-COCO our approach outperforms state-of-the-art methods by 8.8% in caption retrieval, and 11.3% in image retrieval.",A new loss based on relatively hard negatives that achieves state-of-the-art performance in image-caption retrieval.Learning joint embedding of sentences and images using triplet loss that is applied to hardest negatives instead of averaging over all triplets
199,Training Autoencoders by Alternating Minimization,"We present DANTE, a novel method for training neural networks, in particular autoencoders, using the alternating minimization principle.DANTE provides a distinct perspective in lieu of traditional gradient-based backpropagation techniques commonly used to train deep networks.It utilizes an adaptation of quasi-convex optimization techniques to cast autoencoder training as a bi-quasi-convex optimization problem.We show that for autoencoder configurations with both differentiable and non-differentiable activation functions, we can perform the alternations very effectively.DANTE effortlessly extends to networks with multiple hidden layers and varying network configurations.In experiments on standard datasets, autoencoders trained using the proposed method were found to be very promising when compared to those trained using traditional backpropagation techniques, both in terms of training speed, as well as feature extraction and reconstruction performance.","We utilize the alternating minimization principle to provide an effective novel technique to train deep autoencoders.Alternating minimization framework for training autoencoder and encoder-decoder networksThe authors explore an alternating optimization approach for training Auto Encoders, treating each layer as a generalized linear model, and suggest using the stochastic normalized GD as the minimization algorithm in each phase."
200,Transfer Learning for Estimating Causal Effects Using Neural Networks,"We develop new algorithms for estimating heterogeneous treatment effects, combining recent developments in transfer learning for neural networks with insights from the causal inference literature.By taking advantage of transfer learning, we are able to efficiently use different data sources that are related to the same underlying causal mechanisms.We compare our algorithms with those in the extant literature using extensive simulation studies based on large-scale voter persuasion experiments and the MNIST database.Our methods can perform an order of magnitude better than existing benchmarks while using a fraction of the data.","Transfer learning for estimating causal effects using neural networks.Develops algorithms to estimate conditional average treatment effect by auxiliary dataset in different environments, both with and without base learner.The authors propose methods to address a novel task of transfer learning for estimating the CATE function, and evaluate them using a synthetic setting and a real-world experimental dataset.Using neural network regression and comparing transfer learning frameworks to estimate a conditional average treatment effect under string ignorability assumptions"
201,LeMoNADe: Learned Motif and Neuronal Assembly Detection in calcium imaging videos,"Neuronal assemblies, loosely defined as subsets of neurons with reoccurring spatio-temporally coordinated activation patterns, or ""motifs"", are thought to be building blocks of neural representations and information processing.We here propose LeMoNADe, a new exploratory data analysis method that facilitates hunting for motifs in calcium imaging videos, the dominant microscopic functional imaging modality in neurophysiology.Our nonparametric method extracts motifs directly from videos, bypassing the difficult intermediate step of spike extraction.Our technique augments variational autoencoders with a discrete stochastic node, and we show in detail how a differentiable reparametrization and relaxation can be used.An evaluation on simulated data, with available ground truth, reveals excellent quantitative performance.In real video data acquired from brain slices, with no ground truth available, LeMoNADe uncovers nontrivial candidate motifs that can help generate hypotheses for more focused biological investigations.","We present LeMoNADe, an end-to-end learned motif detection method directly operating on calcium imaging videos.This paper proposes a VAE-style model for identifying motifs from calcium imaging videos, relying on Bernouli variables and requires Gumbel-softmax trick for inference."
202,Learning from Noisy Demonstration Sets via Meta-Learned Suitability Assessor,"A noisy and diverse demonstration set may hinder the performances of an agent aiming to acquire certain skills via imitation learning.However, state-of-the-art imitation learning algorithms often assume the optimality of the given demonstration set.In this paper, we address such optimal assumption by learning only from the most suitable demonstrations in a given set.Suitability of a demonstration is estimated by whether imitating it produce desirable outcomes for achieving the goals of the tasks.For more efficient demonstration suitability assessments, the learning agent should be capable of imitating a demonstration as quick as possible, which shares similar spirit with fast adaptation in the meta-learning regime.Our framework, thus built on top of Model-Agnostic Meta-Learning, evaluates how desirable the imitated outcomes are, after adaptation to each demonstration in the set.The resulting assessments hence enable us to select suitable demonstration subsets for acquiring better imitated skills.The videos related to our experiments are available at: https://sites.google.com/view/deepdj","We propose a framework to learn a good policy through imitation learning from a noisy demonstration set via meta-training a demonstration suitability assessor.Contributes a MAML based algorithm to imitation learning which automatically determines if provided demonstrations are ""suitable"".A method for doing imitation learning from a set of demonstrations that includes useless behavior, which selects the useful demonstrations by their provided performance gains at the meta-training time."
203,CausalGAN: Learning Causal Implicit Generative Models with Adversarial Training,"We introduce causal implicit generative models: models that allow sampling from not only the true observational but also the true interventional distributions.We show that adversarial training can be used to learn a CiGM, if the generator architecture is structured based on a given causal graph.We consider the application of conditional and interventional sampling of face images with binary feature labels, such as mustache, young.We preserve the dependency structure between the labels with a given causal graph.We devise a two-stage procedure for learning a CiGM over the labels and the image.First we train a CiGM over the binary labels using a Wasserstein GAN where the generator neural network is consistent with the causal graph between the labels.Later, we combine this with a conditional GAN to generate images conditioned on the binary labels.We propose two new conditional GAN architectures: CausalGAN and CausalBEGAN.We show that the optimal generator of the CausalGAN, given the labels, samples from the image distributions conditioned on these labels.The conditional GAN combined with a trained CiGM for the labels is then a CiGM over the labels and the generated image.We show that the proposed architectures can be used to sample from observational and interventional image distributions, even for interventions which do not naturally occur in the dataset.","We introduce causal implicit generative models, which can sample from conditional and interventional distributions and also propose two new conditional GANs which we use for training them.A method of combining a casual graph, describing the dependency structure of labels with two conditional GAN architechtures that generate images conditioning on the binary labelThe authors address the issue of learning a causal model between image variables and the image itself from observational data, when given a causal structure between image labels."
204, A Matrix Approximation View of NCE that Justifies Self-Normalization,"Self-normalizing discriminative models approximate the normalized probability of a class without having to compute the partition function.This property is useful to computationally-intensive neural network classifiers, as the cost of computing the partition function grows linearly with the number of classes and may become prohibitive.In particular, since neural language models may deal with up to millions of classes, their self-normalization properties received notable attention.Severalrecent studies empirically found that language models, trained using Noise Contrastive Estimation, exhibit self-normalization, but could not explain why.In this study, we provide a theoretical justification to this property by viewingNCE as a low-rank matrix approximation.Our empirical investigation compares NCE to the alternative explicit approach for self-normalizing language models.It also uncovers a surprising negative correlation between self-normalization andperplexity, as well as some regularity in the observed errors that may potentially be used for improving self-normalization algorithms in the future.",We prove that NCE is self-normalized and demonstrate it on datasetsPresents a proof of the self normalization of NCE as a result of being a low-rank matrix approximation of low-rank approximation of the normalized conditional probabilities matrix.This paper considers the problem of self-normalizing models and explains the self-normalizing mechanism by interpreting NCE in terms of matrix factorization.
205,Towards Building Affect sensitive Word Distributions,"Learning word representations from large available corpora relies on the distributional hypothesis that words present in similar contexts tend to have similar meanings.Recent work has shown that word representations learnt in this manner lack sentiment information which, fortunately, can be leveraged using external knowledge.Our work addresses the question: can affect lexica improve the word representations learnt from a corpus?', ""In this work, we propose techniques to incorporate affect lexica, which capture fine-grained information about a word's psycholinguistic and emotional orientation, into the training process of Word2Vec SkipGram, Word2Vec CBOW and GloVe methods using a joint learning approach."", ""We use affect scores from Warriner's affect lexicon to regularize the vector representations learnt from an unlabelled corpus.Our proposed method outperforms previously proposed methods on standard tasks for word similarity detection, outlier detection and sentiment detection.We also demonstrate the usefulness of our approach for a new task related to the prediction of formality, frustration and politeness in corporate communication.",Enriching word embeddings with affect information improves their performance on sentiment prediction tasks.Proposes to use affect lexica to improve word embeddings to outperform the standard Word2vec and Glove.This paper proposes integrating information from a semantic resource quantifying the affect of words into a text-based word embedding algorithm to make language models more reflective of semantic and pragmatic phenomena.This paper introduces modifications the word2vec and GloVe loss functions to incorporate affect lexica to facilitate the learning of affect-sensitive word embeddings.
206,BIGSAGE: unsupervised inductive representation learning of graph via bi-attended sampling and global-biased aggregating,"Different kinds of representation learning techniques on graph have shown significant effect in downstream machine learning tasks.Recently, in order to inductively learn representations for graph structures that is unobservable during training, a general framework with sampling and aggregating was proposed by Hamilton and Ying and had been proved more efficient than transductive methods on fileds like transfer learning or evolving dataset.However, GraphSAGE is uncapable of selective neighbor sampling and lack of memory of known nodes that've been trained.To address these problems, we present an unsupervised method that samples neighborhood information attended by co-occurring structures and optimizes a trainable global bias as a representation expectation for each node in the given graph.Experiments show that our approach outperforms the state-of-the-art inductive and unsupervised methods for representation learning on graphs.","For unsupervised and inductive network embedding, we propose a novel approach to explore most relevant neighbors and preserve previously learnt knowledge of nodes by utilizing bi-attention architecture and introducing global bias, respectivelyThis proposes an extension to GraphSAGE using a global embedding bias matrix in the local aggregating functions and a method to sample interesting nodes."
207,The Importance of Norm Regularization in Linear Graph Embedding: Theoretical Analysis and Empirical Demonstration,"Learning distributed representations for nodes in graphs is a crucial primitive in network analysis with a wide spectrum of applications.Linear graph embedding methods learn such representations by optimizing the likelihood of both positive and negative edges while constraining the dimension of the embedding vectors.We argue that the generalization performance of these methods is not due to the dimensionality constraint as commonly believed, but rather the small norm of embedding vectors.Both theoretical and empirical evidence are provided to support this argument: we prove that the generalization error of these methods can be bounded by limiting the norm of vectors, regardless of the embedding dimension; we show that the generalization performance of linear graph embedding methods is correlated with the norm of embedding vectors, which is small due to the early stopping of SGD and the vanishing gradients.We performed extensive experiments to validate our analysis and showcased the importance of proper norm regularization in practice.",We argue that the generalization of linear graph embedding is not due to the dimensionality constraint but rather the small norm of embedding vectors.The authors show that the generalization error of linear graph embedding methods is bounded by the norm of embedding vectors rather than dimensionality constraintsThe authors propose a theoretical bound on the generalization performance of learning graph embeddings and argue that the norm of the coordinates determines the success of the learnt representation.
208,Quasi-hyperbolic momentum and Adam for deep learning,"Momentum-based acceleration of stochastic gradient descent is widely used in deep learning.We propose the quasi-hyperbolic momentum algorithm as an extremely simple alteration of momentum SGD, averaging a plain SGD step with a momentum step.We describe numerous connections to and identities with other algorithms, and we characterize the set of two-state optimization algorithms that QHM can recover.Finally, we propose a QH variant of Adam called QHAdam, and we empirically demonstrate that our algorithms lead to significantly improved training in a variety of settings, including a new state-of-the-art result on WMT16 EN-DE.We hope that these empirical results, combined with the conceptual and practical simplicity of QHM and QHAdam, will spur interest from both practitioners and researchers.Code is immediately available.","Mix plain SGD and momentum (or do something similar with Adam) for great profit.The paper proposes simple modifications to SGD and Adam, called QH-variants, that can recover the “parent” method and a host of other optimization tricks.A variant of classical momentum which takes a weighted average of momentum and gradient update, and an evaluation of its relationships between other momentum based optimization schemes."
209,Learning to Mix n-Step Returns: Generalizing Lambda-Returns for Deep Reinforcement Learning,"Reinforcement Learning can model complex behavior policies for goal-directed sequential decision making tasks.A hallmark of RL algorithms is Temporal Difference learning: value function for the current state is moved towards a bootstrapped target that is estimated using the next state's value function.lambda-returns define the target of the RL agent as a weighted combination of rewards estimated by using multiple many-step look-aheads.Although mathematically tractable, the use of exponentially decaying weighting of n-step returns based targets in lambda-returns is a rather ad-hoc design choice.Our major contribution is that we propose a generalization of lambda-returns called Confidence-based Autodidactic Returns, wherein the RL agent learns the weighting of the n-step returns in an end-to-end manner.In contrast to lambda-returns wherein the RL agent is restricted to use an exponentially decaying weighting scheme, CAR allows the agent to learn to decide how much it wants to weigh the n-step returns based targets.Our experiments, in addition to showing the efficacy of CAR, also empirically demonstrate that using sophisticated weighted mixtures of multi-step returns considerably outperforms the use of n-step returns.We perform our experiments on the Asynchronous Advantage Actor Critic algorithm in the Atari 2600 domain.","A novel way to generalize lambda-returns by allowing the RL agent to decide how much it wants to weigh each of the n-step returns.Extends the A3C algorithm with lambda returns, and proposes an approach for learning the weights of the returns.The authors present confidence-based autodidactic returns, a Deep learning RL method to adjust the weights of an eligibility vector in TD(lambda)-like value estimation to favour more stable estimates of the state."
210,RETHINKING SELF-DRIVING : MULTI -TASK KNOWLEDGE FOR BETTER GENERALIZATION AND ACCIDENT EXPLANATION ABILITY,"Current end-to-end deep learning driving models have two problems: Poorgeneralization ability of unobserved driving environment when diversity of train-ing driving dataset is limited Lack of accident explanation ability when drivingmodels don’t work as expected.To tackle these two problems, rooted on the be-lieve that knowledge of associated easy task is benificial for addressing difficulttask, we proposed a new driving model which is composed of perception modulefor see and think and driving module for behave, and trained it with multi-taskperception-related basic knowledge and driving knowledge stepwisely. Specifi-cally segmentation map and depth map wereconsidered as what & where and how far knowledge for tackling easier driving-related perception problems before generating final control commands for difficultdriving task.The results of experiments demonstrated the effectiveness of multi-task perception knowledge for better generalization and accident explanation abil-ity.With our method the average sucess rate of finishing most difficult navigationtasks in untrained city of CoRL test surpassed current benchmark method for 15percent in trained weather and 20 percent in untrained weathers.","we proposed a new self-driving model which is composed of perception module for see and think and driving module for behave to acquire better generalization and accident explanation ability.Presents a multitask learning architecture for depth and segmentation map estimation and the driving prediction using a perception module and a driving decision module.A method for a modified end-to-end architecture that has better generalization and explanation ability, is more robust to a different testing setting, and has decoder output that can help with debugging the model.The authors present a multi-task convolutional neural network for end-to-end driving and provide evaluations with the CARLA open source simulator showing better generalization performance in new driving conditions than baselines"
211,MILE: A Multi-Level Framework for Scalable Graph Embedding,"Recently there has been a surge of interest in designing graph embedding methods.Few, if any, can scale to a large-sized graph with millions of nodes due to both computational complexity and memory requirements.In this paper, we relax this limitation by introducing the MultI-Level Embedding framework – a generic methodology allowing contemporary graph embedding methods to scale to large graphs.MILE repeatedly coarsens the graph into smaller ones using a hybrid matching technique to maintain the backbone structure of the graph.It then applies existing embedding methods on the coarsest graph and refines the embeddings to the original graph through a novel graph convolution neural network that it learns.The proposed MILE framework is agnostic to the underlying graph embedding techniques and can be applied to many existing graph embedding methods without modifying them.We employ our framework on several popular graph embedding techniques and conduct embedding for real-world graphs.Experimental results on five large-scale datasets demonstrate that MILE significantly boosts the speed of graph embedding while also often generating embeddings of better quality for the task of node classification.MILE can comfortably scale to a graph with 9 million nodes and 40 million edges, on which existing methods run out of memory or take too long to compute on a modern workstation.",A generic framework to scale existing graph embedding techniques to large graphs.This paper proposes a multi-level embedding framework to be applied on top of existing network embedding methods in order to scale to large scale networks with faster speed.The authors propose a three-stage framework for large-scale graph embedding with improved embedding quality.
212,Unsupervised Adversarial Anomaly  Detection using One-Class Support Vector Machines,"Anomaly detection discovers regular patterns in unlabeled data and identifies the non-conforming data points, which in some cases are the result of malicious attacks by adversaries.Learners such as One-Class Support Vector Machines have been successfully in anomaly detection, yet their performance may degrade significantly in the presence of sophisticated adversaries, who target the algorithm itself by compromising the integrity of the training data.With the rise in the use of machine learning in mission critical day-to-day activities where errors may have significant consequences, it is imperative that machine learning systems are made secure.To address this, we propose a defense mechanism that is based on a contraction of the data, and we test its effectiveness using OCSVMs.The proposed approach introduces a layer of uncertainty on top of the OCSVM learner, making it infeasible for the adversary to guess the specific configuration of the learner.We theoretically analyze the effects of adversarial perturbations on the separating margin of OCSVMs and provide empirical evidence on several benchmark datasets, which show that by carefully contracting the data in low dimensional spaces, we can successfully identify adversarial samples that would not have been identifiable in the original dimensional space.The numerical results show that the proposed method improves OCSVMs performance significantly","A novel method to increase the resistance of OCSVMs against targeted, integrity attacks by selective nonlinear transformations of data to lower dimensions.The authors propose a defense against attacks on the security of one-class SVM based anomaly detectorsThis paper explores how random projections can be used to make OCSVM robust to adversarially perturbed training data."
213,Parametric Information Bottleneck to Optimize Stochastic Neural Networks,"In this paper, we present a layer-wise learning of stochastic neural networks in an information-theoretic perspective.In each layer of an SNN, the compression and the relevance are defined to quantify the amount of information that the layer contains about the input space and the target space, respectively.We jointly optimize the compression and the relevance of all parameters in an SNN to better exploit the neural network's representation.Previously, the Information Bottleneck framework extracts relevant information for a target variable.Here, we propose Parametric Information Bottleneck for a neural network by utilizing its model parameters explicitly to approximate the compression and the relevance.We show that, as compared to the maximum likelihood estimate principle, PIBs : improve the generalization of neural networks in classification tasks, push the representation of neural networks closer to the optimal information-theoretical representation in a faster manner. ","Learning a better neural networks' representation with Information Bottleneck principle"", 'Proposes a learning method based on the information bottleneck framework, where hidden layers of deep nets compress the input X while maintaining sufficient information to predict the output Y.This paper presents a new way of training stochastic neural network following an information relevance/compression framework similar to the Information Bottleneck."
214,Directing Generative Networks with Weighted Maximum Mean Discrepancy,"The maximum mean discrepancy between two probability measures Pand Q is a metric that is zero if and only if all moments of the two measuresare equal, making it an appealing statistic for two-sample tests.Given i.i.d. samplesfrom P and Q, Gretton et al. show that we can construct an unbiasedestimator for the square of the MMD between the two distributions.If P is adistribution of interest and Q is the distribution implied by a generative neuralnetwork with stochastic inputs, we can use this estimator to train our neural network.However, in practice we do not always have i.i.d. samples from our targetof interest.Data sets often exhibit biases—for example, under-representation ofcertain demographics—and if we ignore this fact our machine learning algorithmswill propagate these biases.Alternatively, it may be useful to assume our data hasbeen gathered via a biased sample selection mechanism in order to manipulateproperties of the estimating distribution Q.In this paper, we construct an estimator for the MMD between P and Q when weonly have access to P via some biased sample selection mechanism, and suggestmethods for estimating this sample selection mechanism when it is not alreadyknown.We show that this estimator can be used to train generative neural networkson a biased data sample, to give a simulator that reverses the effect of thatbias.","We propose an estimator for the maximum mean discrepancy, appropriate when a target distribution is only accessible via a biased sample selection procedure, and show that it can be used in a generative network to correct for this bias.Proposes an importance-weighted estimator of the MMD to estimate the MMD between distributions based on samples biased according to a known or estimated unknown scheme.The authors address the problem of sample selection bias in MMD-GANs and propose an estimate of the MMD between two distributions using weighted maximum mean discrepancy.This paper presents a modification of the objective used to train generative networks with an MMD adversary "
215,Efficient Exploration through Bayesian   Deep Q-Networks,"We propose Bayesian Deep Q-Network , a practical Thompson sampling based Reinforcement Learning Algorithm.Thompson sampling allows for targeted exploration in high dimensions through posterior sampling but is usually computationally expensive.We address this limitation by introducing uncertainty only at the output layer of the network through a Bayesian Linear Regression model, which can be trained with fast closed-form updates and its samples can be drawn efficiently through the Gaussian distribution.We apply our method to a wide range of Atari Arcade Learning Environments.Since BDQN carries out more efficient exploration, it is able to reach higher rewards substantially faster than a key baseline, DDQN.",Using Bayesian regression to estimate the posterior over Q-functions and deploy Thompson Sampling as a targeted exploration strategy with efficient trade-off the exploration and exploitationThe authors propose a new algorithm for exploration in Deep RL where they apply Bayesian linear regression with features from the last layer of a DQN network to estimate the Q function for each action.The authors describe how to use Bayesian neural networks with Thompson sampling for efficient exploration in q-learning and propose an approach that outperforms epsilon-greedy exploration approaches.
216,PolyCNN: Learning Seed Convolutional Filters,"In this work, we propose the polynomial convolutional neural network, as a new design of a weight-learning efficient variant of the traditional CNN.The biggest advantage of the PolyCNN is that at each convolutional layer, only one convolutional filter is needed for learning the weights, which we call the seed filter, and all the other convolutional filters are the polynomial transformations of the seed filter, which is termed as an early fan-out.Alternatively, we can also perform late fan-out on the seed filter response to create the number of response maps needed to be input into the next layer.Both early and late fan-out allow the PolyCNN to learn only one convolutional filter at each layer, which can dramatically reduce the model complexity by saving 10x to 50x parameters during learning.While being efficient during both training and testing, the PolyCNN does not suffer performance due to the non-linear polynomial expansion which translates to richer representational power within the convolutional layers.By allowing direct control over model complexity, PolyCNN provides a flexible trade-off between performance and efficiency.We have verified the on-par performance between the proposed PolyCNN and the standard CNN on several visual datasets, such as MNIST, CIFAR-10, SVHN, and ImageNet.","PolyCNN only needs to learn one seed convolutional filter at each layer. This is an efficient variant of traditional CNN, with on-par performance.Attempts at reducing the number of CNN model parameters by using the polynomial transformation of filters to create blow-up the filter responses.The authors propose a weight sharing architecture for reducing the number of convolutional neural network parameters with seed filters"
217,Kernel Change-point Detection with Auxiliary Deep Generative Models,"Detecting the emergence of abrupt property changes in time series is a challenging problem.Kernel two-sample test has been studied for this task which makes fewer assumptions on the distributions than traditional parametric approaches.However, selecting kernels is non-trivial in practice.Although kernel selection for the two-sample test has been studied, the insufficient samples in change point detection problem hinder the success of those developed kernel selection algorithms.In this paper, we propose KL-CPD, a novel kernel learning framework for time series CPD that optimizes a lower bound of test power via an auxiliary generative model.With deep kernel parameterization, KL-CPD endows kernel two-sample test with the data-driven kernel to detect different types of change-points in real-world applications.The proposed approach significantly outperformed other state-of-the-art methods in our comparative evaluation of benchmark datasets and simulation studies.","In this paper, we propose KL-CPD, a novel kernel learning framework for time series CPD that optimizes a lower bound of test power via an auxiliary generative model as a surrogate to the abnormal distribution. Describes a novel approach to optimising the choice of kernel towards increased testing power and shown to offer improvements over alternatives."
218,Cluster-based Warm-Start Nets,"Theories in cognitive psychology postulate that humans use similarity as a basisfor object categorization.However, work in image classification generally as-sumes disjoint and equally dissimilar classes to achieve super-human levels ofperformance on certain datasets.In our work, we adapt notions of similarity usingweak labels over multiple hierarchical levels to boost classification performance.Instead of pitting clustering directly against classification, we use a warm-startbased evaluation to explicitly provide value to a clustering representation by itsability to aid classification.We evaluate on CIFAR10 and a fine-grained classifi-cation dataset to show improvements in performance with the procedural additionof intermediate losses and weak labels based on multiple hierarchy levels.Further-more, we show that pretraining AlexNet on hierarchical weak labels in conjunc-tion with intermediate losses outperforms a classification baseline by over 17% ona subset of Birdsnap dataset.Finally, we show improvement over AlexNet trainedusing ImageNet pre-trained weights as initializations which further supports ourclaim of the importance of similarity.",Cluster before you classify; using weak labels to improve classification Proposes using a clustering based loss function at multiple levels of a deepnet as well as using hierarchical structure of the label space to train better representations.This paper uses hierarchical label information to impose additional losses on intermediate representations in neural network training.
219,Regret Minimization for Partially Observable Deep Reinforcement Learning,"Deep reinforcement learning algorithms that estimate state and state-action value functions have been shown to be effective in a variety of challenging domains, including learning control strategies from raw image pixels.However, algorithms that estimate state and state-action value functions typically assume a fully observed state and must compensate for partial or non-Markovian observations by using finite-length frame-history observations or recurrent networks.In this work, we propose a new deep reinforcement learning algorithm based on counterfactual regret minimization that iteratively updates an approximation to a cumulative clipped advantage function and is robust to partially observed state.We demonstrate that on several partially observed reinforcement learning tasks, this new class of algorithms can substantially outperform strong baseline methods: on Pong with single-frame observations, and on the challenging Doom and Minecraft first-person navigation benchmarks.","Advantage-based regret minimization is a new deep reinforcement learning algorithm that is particularly effective on partially observable tasks, such as 1st person navigation in Doom and Minecraft.This paper introduces the concepts of counterfactual regret minimization in the field of Deep RL and an algorithm called ARM which can deal with partial observability better.The paper provides a game-theoretic inspired variant of policy-gradient algorithm based on the idea of counter-factual regret minimization and claims that the approach can deal with the partial observable domain better than standard methods."
220,TENSOR RING NETS ADAPTED DEEP MULTI-TASK LEARNING,"Recent deep multi-task learning has been witnessed its success in alleviating data scarcity of some task by utilizing domain-specific knowledge from related tasks.Nonetheless, several major issues of deep MTL, including the effectiveness of sharing mechanisms, the efficiency of model complexity and the flexibility of network architectures, still remain largely unaddressed.To this end, we propose a novel generalized latent-subspace based knowledge sharing mechanism for linking task-specific models, namely tensor ring multi-task learning.TRMTL has a highly compact representation, and it is very effective in transferring task-invariant knowledge while being super flexible in learning task-specific features, successfully mitigating the dilemma of both negative-transfer in lower layers and under-transfer in higher layers.Under our TRMTL, it is feasible for each task to have heterogenous input data dimensionality or distinct feature sizes at different hidden layers.Experiments on a variety of datasets demonstrate our model is capable of significantly improving each single task’s performance, particularly favourable in scenarios where some of the tasks have insufficient data.","a deep multi-task learning model adapting tensor ring representationA variant of tensor ring formulation for multi-task learning by sharing some of the TT cores for learning ""common task"" while learning individual TT cores for each separate task"
221,Attentive Neural Processes,"Neural Processes approach regression by learning to map a context set of observed input-output pairs to a distribution over regression functions.Each function models the distribution of the output given an input, conditioned on the context.NPs have the benefit of fitting observed data efficiently with linear complexity in the number of context input-output pairs, and can learn a wide family of conditional distributions; they learn predictive distributions conditioned on context sets of arbitrary size.Nonetheless, we show that NPs suffer a fundamental drawback of underfitting, giving inaccurate predictions at the inputs of the observed data they condition on.We address this issue by incorporating attention into NPs, allowing each input location to attend to the relevant context points for the prediction.We show that this greatly improves the accuracy of predictions, results in noticeably faster training, and expands the range of functions that can be modelled.","A model for regression that learns conditional distributions of a stochastic process, by incorporating attention into Neural Processes.Proposes to resolve the issue of underfitting in the neural process method by adding an attention mechanism to the deterministic path.An extension to the framework of Neural Processes that adds an attention-based conditioning mechanism, allowing the model to better capture dependencies in the conditioning set.The authors extend neural processes by incorporating self-attention for enriching the features of the context points and cross-attention for producing a query-specific representation. They resolve the underfitting problem of NPs and show ANPs to converge better and faster than NPs."
222,Pixel Deconvolutional Networks,"Deconvolutional layers have been widely used in a variety of deepmodels for up-sampling, including encoder-decoder networks forsemantic segmentation and deep generative models for unsupervisedlearning.One of the key limitations of deconvolutional operationsis that they result in the so-called checkerboard problem.This iscaused by the fact that no direct relationship exists among adjacentpixels on the output feature map.To address this problem, wepropose the pixel deconvolutional layer to establishdirect relationships among adjacent pixels on the up-sampled featuremap.Our method is based on a fresh interpretation of the regulardeconvolution operation.The resulting PixelDCL can be used toreplace any deconvolutional layer in a plug-and-play manner withoutcompromising the fully trainable capabilities of original models.The proposed PixelDCL may result in slight decrease in efficiency,but this can be overcome by an implementation trick.Experimentalresults on semantic segmentation demonstrate that PixelDCL canconsider spatial features such as edges and shapes and yields moreaccurate segmentation outputs than deconvolutional layers.When usedin image generation tasks, our PixelDCL can largely overcome thecheckerboard problem suffered by regular deconvolution operations.","Solve checkerboard problem in Deconvolutional layer by building dependencies between pixelsThis work proposes pixel deconvolutional layers for convolutional neural networks as a way to alleviate the checkerboard effect.A novel technique to generalize deconvolution operations used in standard CNN architectures, which proposes doing sequential prediction of adjacent pixel features, resulting in more spatially smooth outputs for deconvolution layers."
223,Variational Network Quantization,"In this paper, the preparation of a neural network for pruning and few-bit quantization is formulated as a variational inference problem.To this end, a quantizing prior that leads to a multi-modal, sparse posterior distribution over weights, is introduced and a differentiable Kullback-Leibler divergence approximation for this prior is derived.After training with Variational Network Quantization, weights can be replaced by deterministic quantization values with small to negligible loss of task accuracy.The method does not require fine-tuning after quantization.Results are shown for ternary quantization on LeNet-5 and DenseNet.","We quantize and prune neural network weights using variational Bayesian inference with a multi-modal, sparsity inducing prior.Proposes to use a mixture of continuous spike propto 1/abs as prior for a Bayesian neural network and demonstrates the good performance with relatively sparsified convnets for minist and cifar-10.This paper presents a variational Bayesian approach for quantising neural network weights to ternary values post-training in a principled way."
224,Progressive Weight Pruning Of Deep Neural Networks Using ADMM,"Deep neural networks although achieving human-level performance in many domains, have very large model size that hinders their broader applications on edge computing devices.Extensive research work have been conducted on DNN model compression or pruning.However, most of the previous work took heuristic approaches.This work proposes a progressive weight pruning approach based on ADMM, a powerful technique to deal with non-convex optimization problems with potentially combinatorial constraints.Motivated by dynamic programming, the proposed method reaches extremely high pruning rate by using partial prunings with moderate pruning rates.Therefore, it resolves the accuracy degradation and long convergence time problems when pursuing extremely high pruning ratios.It achieves up to 34× pruning rate for ImageNet dataset and 167× pruning rate for MNIST dataset, significantly higher than those reached by the literature work.Under the same number of epochs, the proposed method also achieves faster convergence and higher compression rates.The codes and pruned DNN models are released in the anonymous link bit.ly/2zxdlss.","We implement a DNN weight pruning approach that achieves the highest pruning rates.This paper focuses on weight pruning for neural network compression, achiving 30x compression rate for AlexNet and VGG for ImageNet.A progressive pruning technique which imposes structural sparsity constraint on the weight parameter and rewrites the optimization as an ADMM framework, achieving higher accurancy than projected gradient descent."
225,Interpolation-Prediction Networks for Irregularly Sampled Time Series,"In this paper, we present a new deep learning architecture for addressing the problem of supervised learning with sparse and irregularly sampled multivariate time series.The architecture is based on the use of a semi-parametric interpolation network followed by the application of a prediction network.The interpolation network allows for information to be shared across multiple dimensions of a multivariate time series during the interpolation stage, while any standard deep learning model can be used for the prediction network.This work is motivated by the analysis of physiological time series data in electronic health records, which are sparse, irregularly sampled, and multivariate.We investigate the performance of this architecture on both classification and regression tasks, showing that our approach outperforms a range of baseline and recently proposed models.","This paper presents a new deep learning architecture for addressing the problem of supervised learning with sparse and irregularly sampled multivariate time series.Proposes a framework for making predictions on sparse, irregularly sampled time-series data using an interpolation module that models the missing values in using smooth interpolation, non-smooth interpolation, and intensity. Solves the problem of supervised learning with sparse and irregularly sampled multivariate time series using a semi-parametric interpolation network followed by a prediction network."
226,A Hierarchical Model for Device Placement,"We introduce a hierarchical model for efficient placement of computational graphs onto hardware devices, especially in heterogeneous environments with a mixture of CPUs, GPUs, and other computational devices.Our method learns to assign graph operations to groups and to allocate those groups to available devices.The grouping and device allocations are learned jointly.The proposed method is trained with policy gradient and requires no human intervention.Experiments with widely-usedcomputer vision and natural language models show that our algorithm can find optimized, non-trivial placements for TensorFlow computational graphs with over 80,000 operations.In addition, our approach outperforms placements by humanexperts as well as a previous state-of-the-art placement method based on deep reinforcement learning.Our method achieves runtime reductions of up to 60.6% per training step when applied to models such as Neural Machine Translation.","We introduce a hierarchical model for efficient, end-to-end placement of computational graphs onto hardware devices.Proposes to jointly learn groups of operators to colocate and to place learned groups on devices to distribute operations for deep learning via reinforcement learning. The authors purpose a fully connect network to replace the co-location step in an auto-placement method proposed to accelerate a TensorFlow model's runtime."", 'Proposes a device placement algorithm to place operations of tensorflow on devices."
227,Understanding image motion with group representations ,"Motion is an important signal for agents in dynamic environments, but learning to represent motion from unlabeled video is a difficult and underconstrained problem.We propose a model of motion based on elementary group properties of transformations and use it to train a representation of image motion.While most methods of estimating motion are based on pixel-level constraints, we use these group properties to constrain the abstract representation of motion itself.We demonstrate that a deep neural network trained using this method captures motion in both synthetic 2D sequences and real-world sequences of vehicle motion, without requiring any labels.Networks trained to respect these constraints implicitly identify the image characteristic of motion in different sequence types.In the context of vehicle motion, this method extracts information useful for localization, tracking, and odometry.Our results demonstrate that this representation is useful for learning motion in the general setting where explicit labels are difficult to obtain.","We propose of method of using group properties to learn a representation of motion without labels and demonstrate the use of this method for representing 2D and 3D motion.Proposes to learn the rigid motion group from a latent representation of image sequences without the need for explicit labels and experimentally demonstrates method on sequences of MINST digits and the KITTI dataset.This paper proposes an approach for learning video motion features in an unsupervised manner, using constraints to optimize the neural network to produce features that can be used to regress odometry."
228,Continuous Convolutional Neural Networks for Image Classification,"This paper introduces the concept of continuous convolution to neural networks and deep learning applications in general.Rather than directly using discretized information, input data is first projected into a high-dimensional Reproducing Kernel Hilbert Space, where it can be modeled as a continuous function using a series of kernel bases.We then proceed to derive a closed-form solution to the continuous convolution operation between two arbitrary functions operating in different RKHS.Within this framework, convolutional filters also take the form of continuous functions, and the training procedure involves learning the RKHS to which each of these filters is projected, alongside their weight parameters.This results in much more expressive filters, that do not require spatial discretization and benefit from properties such as adaptive support and non-stationarity.Experiments on image classification are performed, using classical datasets, with results indicating that the proposed continuous convolutional neural network is able to achieve competitive accuracy rates with far fewer parameters and a faster convergence rate.",This paper proposes a novel convolutional layer that operates in a continuous Reproducing Kernel Hilbert Space.Projecting examples into an RK Hilbert space and performing convolution and filtering into that space.This paper formulates a variant of convolutional neural networks which models both activations and filters as continuous functions composed from kernel bases
229,ImageNet-trained CNNs are biased towards texture; increasing shape bias improves accuracy and robustness,"Convolutional Neural Networks are commonly thought to recognise objects by learning increasingly complex representations of object shapes.Some recent studies suggest a more important role of image textures.We here put these conflicting hypotheses to a quantitative test by evaluating CNNs and human observers on images with a texture-shape cue conflict.We show that ImageNet-trained CNNs are strongly biased towards recognising textures rather than shapes, which is in stark contrast to human behavioural evidence and reveals fundamentally different classification strategies.We then demonstrate that the same standard architecture that learns a texture-based representation on ImageNet is able to learn a shape-based representation instead when trained on 'Stylized-ImageNet', a stylized version of ImageNet.This provides a much better fit for human behavioural performance in our well-controlled psychophysical lab setting and comes with a number of unexpected emergent benefits such as improved object detection performance and previously unseen robustness towards a wide range of image distortions, highlighting advantages of a shape-based representation.","ImageNet-trained CNNs are biased towards object texture (instead of shape like humans). Overcoming this major difference between human and machine vision yields improved detection performance and previously unseen robustness to image distortions.Using image stylizaton to augment training data for ImageNet-trained CNNs to make resulting networks appear more aligned with human judgementsThis paper studies CNNs like AlexNet, VGG, GoogleNet, and ResNet50, shows these models are biased towards texture when trained on ImageNet, and proposes a new ImageNet dataset."
230,Variational Autoencoders with implicit priors for short-duration text-independent speaker verification,"In this work, we exploited different strategies to provide prior knowledge to commonly used generative modeling approaches aiming to obtain speaker-dependent low dimensional representations from short-duration segments of speech data, making use of available information of speaker identities.Namely, convolutional variational autoencoders are employed, and statistics of its learned posterior distribution are used as low dimensional representations of fixed length short-duration utterances.In order to enforce speaker dependency in the latent layer, we introduced a variation of the commonly used prior within the variational autoencoders framework, i.e. the model is simultaneously trained for reconstruction of inputs along with a discriminative task performed on top of latent layers outputs.The effectiveness of both triplet loss minimization and speaker recognition are evaluated as implicit priors on the challenging cross-language NIST SRE 2016 setting and compared against fully supervised and unsupervised baselines.",We evaluate the effectiveness of having auxiliary discriminative tasks performed on top of statistics of the posterior distribution learned by variational autoencoders to enforce speaker dependency.Propose an autoencoder model to learn a representation for speaker verification using short-duration analysis windows.A modified version of the variational autoencoder model that tackles the speaker recognition problem in the context of short-duration segments
231,Debiasing Evidence Approximations: On Importance-weighted Autoencoders and Jackknife Variational Inference,"The importance-weighted autoencoder approach of Burda et al. defines a sequence of increasingly tighter bounds on the marginal likelihood of latent variable models.Recently, Cremer et al. reinterpreted the IWAE bounds as ordinary variational evidence lower bounds applied to increasingly accurate variational distributions.In this work, we provide yet another perspective on the IWAE bounds.We interpret each IWAE bound as a biased estimator of the true marginal likelihood where for the bound defined on samples we show the bias to be of order O.In our theoretical analysis of the IWAE objective we derive asymptotic bias and variance expressions.Based on this analysis we develop jackknife variational inference,a family of bias-reduced estimators reducing the bias to for any given m < K while retaining computational efficiency.Finally, we demonstrate that JVI leads to improved evidence estimates in variational autoencoders.We also report first results on applying JVI to learning variational autoencoders.Our implementation is available at https://github.com/Microsoft/jackknife-variational-inference","Variational inference is biased, let's debias it."", 'Introduces jackknife variational inference, a method for debiasing Monte Carlo objectives such as the importance weighted auto-encoder.The authors analyze the bias and variance of the IWAE bound and derive a jacknife approach to estimate moments as a way to debias IWAE for finite importance weighted samples."
232,Tactical Decision Making for Lane Changing with Deep Reinforcement Learning,"In this paper, we consider the problem of autonomous lane changing for self driving vehicles in a multi-lane, multi-agent setting.We present a framework that demonstrates a more structured and data efficient alternative to end-to-end complete policy learning on problems where the high-level policy is hard to formulate using traditional optimization or rule based methods but well designed low-level controllers are available.Our framework uses deep reinforcement learning solely to obtain a high-level policy for tactical decision making, while still maintaining a tight integration with the low-level controller, thus getting the best of both worlds.We accomplish this with Q-masking, a technique with which we are able to incorporate prior knowledge, constraints, and information from a low-level controller, directly in to the learning process thereby simplifying the reward function and making learning faster and data efficient.We provide preliminary results in a simulator and show our approach to be more efficient than a greedy baseline, and more successful and safer than human driving.","A framework that provides a policy for autonomous lane changing by learning to make high-level tactical decisions with deep reinforcement learning, and maintaining a tight integration with a low-level controller to take low-level actions.Considers the problem of autonomous lane changing for self-driving cars in multi-lane multi-agent slot car setting, proposes a new learning strategy Q-masking - coupling a defined low level controller with a high level tactical decision making policy.This paper proposes a deep Q-learning approach to the problem of lane change using ""Q-masking,"" which reduces the action space according to contraints or prior knowledge.Authors propose a method which uses a Q-learning-based high-level policy that is combined with a contextual mask derived from safety-contraints and low-level controllers, which disable certain actions from being selectable at certain states. "
233,Neural Graph Evolution: Towards Efficient Automatic Robot Design,"Despite the recent successes in robotic locomotion control, the design of robot relies heavily on human engineering.Automatic robot design has been a long studied subject, but the recent progress has been slowed due to the large combinatorial search space and the difficulty in evaluating the found candidates.To address the two challenges, we formulate automatic robot design as a graph search problem and perform evolution search in graph space.We propose Neural Graph Evolution, which performs selection on current candidates and evolves new ones iteratively.Different from previous approaches, NGE uses graph neural networks to parameterize the control policies, which reduces evaluation cost on new candidates with the help of skill transfer from previously evaluated designs.In addition, NGE applies Graph Mutation with Uncertainty by incorporating model uncertainty, which reduces the search space by balancing exploration and exploitation.We show that NGE significantly outperforms previous methods by an order of magnitude.As shown in experiments, NGE is the first algorithm that can automatically discover kinematically preferred robotic graph structures, such as a fish with two symmetrical flat side-fins and a tail, or a cheetah with athletic front and back legs.Instead of using thousands of cores for weeks, NGE efficiently solves searching problem within a day on a single 64 CPU-core Amazon EC2machine.","Automatic robotic design search with graph neural networksProposes an approach for automatic robot design based on Neural graph evolution. The experiments demonstrate that optimizing both controller and hardware is better than optimizing just the controller.The authors propose a scheme based on a graph representation of the robot structure, and a graph-neural-network as controllers to optimize robot structures, combined with their controllers. "
234,GraphVAE: Towards Generation of Small Graphs Using Variational Autoencoders,"Deep learning on graphs has become a popular research topic with many applications.However, past work has concentrated on learning graph embedding tasks only, which is in contrast with advances in generative models for images and text.Is it possible to transfer this progress to the domain of graphs?We propose to sidestep hurdles associated with linearization of such discrete structures by having a decoder output a probabilistic fully-connected graph of a predefined maximum size directly at once.Our method is formulated as a variational autoencoder.We evaluate on the challenging task of conditional molecule generation.","We demonstate an autoencoder for graphs.Learning to generate graphs using deep learning methods in ""one shot"", directly outputting node and edge existence probabilities, and node attribute vectors.A variational auto encoder to generate graphs"
235,Towards Binary-Valued Gates for Robust LSTM Training ,"Long Short-Term Memory is one of the most widely used recurrent structures in sequence modeling.Its goal is to use gates to control the information flow in the recurrent computations, although its practical implementation based on soft gates only partially achieves this goal and is easy to overfit.In this paper, we propose a new way for LSTM training, which pushes the values of the gates towards 0 or 1.By doing so, we can better control the information flow: the gates are mostly open or closed, instead of in a middle state; and avoid overfitting to certain extent: the gates operate at their flat regions, which is shown to correspond to better generalization ability.However, learning towards discrete values of the gates is generally difficult.To tackle this challenge, we leverage the recently developed Gumbel-Softmax trick from the field of variational methods, and make the model trainable with standard backpropagation.Experimental results on language modeling and machine translation show that the values of the gates generated by our method are more reasonable and intuitively interpretable, and our proposed method generalizes better and achieves better accuracy on test sets in all tasks.Moreover, the learnt models are not sensitive to low-precision approximation and low-rank approximation of the gate parameters due to the flat loss surface.","We propose a new algorithm for LSTM training by learning towards binary-valued gates which we shown has many nice properties.Propose a new ""gate"" function for LSTM to enable the values of the gates towards 0 or 1. The paper aims to push LSTM gates to be binary by employing the recent Gumbel-Softmax trick to obtain end-to-end trainable categorical distribution."
236,THE EFFECTIVENESS OF A TWO-LAYER NEURAL NETWORK FOR RECOMMENDATIONS,"We present a personalized recommender system using neural network for recommendingproducts, such as eBooks, audio-books, Mobile Apps, Video and Music.It produces recommendations based on customer’s implicit feedback history suchas purchases, listens or watches.Our key contribution is to formulate recommendationproblem as a model that encodes historical behavior to predict the futurebehavior using soft data split, combining predictor and auto-encoder models.Weintroduce convolutional layer for learning the importance of the purchasesdepending on their purchase date and demonstrate that the shape of the timedecay function can be well approximated by a parametrical function.We presentoffline experimental results showing that neural networks with two hidden layerscan capture seasonality changes, and at the same time outperform other modelingtechniques, including our recommender in production.Most importantly, wedemonstrate that our model can be scaled to all digital categories, and we observesignificant improvements in an online A/B test.We also discuss key enhancementsto the neural network model and describe our production pipeline.Finallywe open-sourced our deep learning library which supports multi-gpu model paralleltraining.This is an important feature in building neural network based recommenderswith large dimensionality of input and output data.",Improving recommendations using time sensitive modeling with neural networks in multiple product categories on a retail websiteThe paper proposes a new neural network based method for recommendation.The authors describe a procedure of building their production recommender system from scratch and integrate time decay of purchases into the learning framework.
237,Task-GAN for Improved GAN based Image Restoration,"Deep Learning algorithms based on Generative Adversarial Network have demonstrated great potentials in computer vision tasks such as image restoration.Despite the rapid development of image restoration algorithms using DL and GANs, image restoration for specific scenarios, such as medical image enhancement and super-resolved identity recognition, are still facing challenges.How to ensure visually realistic restoration while avoiding hallucination or mode- collapse?How to make sure the visually plausible results do not contain hallucinated features jeopardizing downstream tasks such as pathology identification and subject identification?Here we propose to resolve these challenges by coupling the GAN based image restoration framework with another task-specific network.With medical imaging restoration as an example, the proposed model conducts additional pathology recognition/classification task to ensure the preservation of detailed structures that are important to this task.Validated on multiple medical datasets, we demonstrate the proposed method leads to improved deep learning based image restoration while preserving the detailed structure and diagnostic features.Additionally, the trained task network show potentials to achieve super-human level performance in identifying pathology and diagnosis.Further validation on super-resolved identity recognition tasks also show that the proposed method can be generalized for diverse image restoration tasks.","Couple the GAN based image restoration framework with another task-specific network to generate realistic image while preserving task-specific features.A novel method of Task-GAN of image coupling that couples GAN and a task-specific network, which alleviates to avoid hallucination or mode collapse.The authors propose to augment GAN-based image restoration with another task-specific branch, such as classification tasks, for further improvement."
238,Deep Autoencoding Gaussian Mixture Model for Unsupervised Anomaly Detection,"Unsupervised anomaly detection on multi- or high-dimensional data is of great importance in both fundamental machine learning research and industrial applications, for which density estimation lies at the core.Although previous approaches based on dimensionality reduction followed by density estimation have made fruitful progress, they mainly suffer from decoupled model learning with inconsistent optimization goals and incapability of preserving essential information in the low-dimensional space.In this paper, we present a Deep Autoencoding Gaussian Mixture Model for unsupervised anomaly detection.Our model utilizes a deep autoencoder to generate a low-dimensional representation and reconstruction error for each input data point, which is further fed into a Gaussian Mixture Model.Instead of using decoupled two-stage training and the standard Expectation-Maximization algorithm, DAGMM jointly optimizes the parameters of the deep autoencoder and the mixture model simultaneously in an end-to-end fashion, leveraging a separate estimation network to facilitate the parameter learning of the mixture model.The joint optimization, which well balances autoencoding reconstruction, density estimation of latent representation, and regularization, helps the autoencoder escape from less attractive local optima and further reduce reconstruction errors, avoiding the need of pre-training.Experimental results on several public benchmark datasets show that, DAGMM significantly outperforms state-of-the-art anomaly detection techniques, and achieves up to 14% improvement based on the standard F1 score.","An end-to-end trained deep neural network that leverages Gaussian Mixture Modeling to perform density estimation and unsupervised anomaly detection in a low-dimensional space learned by deep autoencoder.The paper presents a joint deep learning framework for dimension reduction-clustering, leads to competitive anomaly detection.A new technique for anomaly detection where the dimension reduction and density estimation steps are jointly optimized."
239,Projective Subspace Networks For Few-Shot Learning,"Generalization from limited examples, usually studied under the umbrella of meta-learning, equips learning techniques with the ability to adapt quickly in dynamical environments and proves to be an essential aspect of lifelong learning.In this paper, we introduce the Projective Subspace Networks, a deep learning paradigm that learns non-linear embeddings from limited supervision.In contrast to previous studies, the embedding in PSN deems samples of a given class to form an affine subspace.We will show that such modeling leads to robust solutions, yielding competitive results on supervised and semi-supervised few-shot classification.Moreover, our PSN approach has the ability of end-to-end learning.In contrast to previous works, our projective subspace can be thought of as a richer representation capturing higher-order information datapoints for modeling new concepts.",We proposed Projective Subspace Networks for few-shot and semi-supervised few-shot learningThis paper proposes a new embedding-based approach for the problem of few-shot learning and an extension to this model to the semi-supervised few-shot learning setting.New method for fully and semi-supervised few-shot classification based on learning a general embedding and then learning a subspace of it for each class
240,Contingency-Aware Exploration in Reinforcement Learning,"This paper investigates whether learning contingency-awareness and controllable aspects of an environment can lead to better exploration in reinforcement learning.To investigate this question, we consider an instantiation of this hypothesis evaluated on the Arcade Learning Element.In this study, we develop an attentive dynamics model that discovers controllable elements of the observations, which are often associated with the location of the character in Atari games.The ADM is trained in a self-supervised fashion to predict the actions taken by the agent.The learned contingency information is used as a part of the state representation for exploration purposes.We demonstrate that combining actor-critic algorithm with count-based exploration using our representation achieves impressive results on a set of notoriously challenging Atari games due to sparse rewards.For example, we report a state-of-the-art score of >11,000 points on Montezuma's Revenge without using expert demonstrations, explicit high-level information, or supervisory data.Our experiments confirm that contingency-awareness is indeed an extremely powerful concept for tackling exploration problems in reinforcement learning and opens up interesting research questions for further investigations.","We investigate contingency-awareness and controllable aspects in exploration and achieve state-of-the-art performance on Montezuma's Revenge without expert demonstrations."", 'This paper investigates the problem of extracting a meaningful state representation to help with exploration when confronted with a sparse reward task by identifying controllable (learned) features of the stateThis paper proposes the novel idea of using contingency awareness to aid exploration in sparse-reward reinforcement learning tasks, obtaining state of the art results."
241,DNA-GAN: Learning Disentangled Representations from Multi-Attribute Images,"Disentangling factors of variation has always been a challenging problem in representation learning.Existing algorithms suffer from many limitations, such as unpredictable disentangling factors, bad quality of generated images from encodings, lack of identity information, etc.In this paper, we proposed a supervised algorithm called DNA-GAN trying to disentangle different attributes of images.The latent representations of images are DNA-like, in which each individual piece represents an independent factor of variation.By annihilating the recessive piece and swapping a certain piece of two latent representations, we obtain another two different representations which could be decoded into images.In order to obtain realistic images and also disentangled representations, we introduced the discriminator for adversarial training.Experiments on Multi-PIE and CelebA datasets demonstrate the effectiveness of our method and the advantage of overcoming limitations existing in other methods.","We proposed a supervised algorithm, DNA-GAN, to disentangle multiple attributes of images.This paper investigates the problem of attribute-conditioned image generation using generative adversarial networks, and proposes to generate images from attribute and latent code as high-level representation.This paper proposed a new method to disentangle different attributes of images using a novel DNA structure GAN"
242,Invariant-equivariant representation learning for multi-class data,"Representations learnt through deep neural networks tend to be highly informative, but opaque in terms of what information they learn to encode.We introduce an approach to probabilistic modelling that learns to represent data with two separate deep representations: an invariant representation that encodes the information of the class from which the data belongs, and an equivariant representation that encodes the symmetry transformation defining the particular data point within the class manifold.This approach to representation learning is conceptually transparent, easy to implement, and in-principle generally applicable to any data comprised of discrete classes of continuous distributions.We demonstrate qualitatively compelling representation learning and competitive quantitative performance, in both supervised and semi-supervised settings, versus comparable modelling approaches in the literature with little fine tuning.",This paper presents a novel latent-variable generative modelling technique that enables the representation of global information into one latent variable and local information into another latent variable.The paper presents a VAE that uses labels to separate the learned representation into an invariant and a covariant part.
243,Active Learning for Convolutional Neural Networks: A Core-Set Approach,"Convolutional neural networks have been successfully applied to many recognition and learning tasks using a universal recipe; training a deep model on a very large dataset of supervised examples.However, this approach is rather restrictive in practice since collecting a large set of labeled images is very expensive.One way to ease this problem is coming up with smart ways for choosing images to be labelled from a very large collection.Our empirical study suggests that many of the active learning heuristics in the literature are not effective when applied to CNNs when applied in batch setting.Inspired by these limitations, we define the problem of active learning as core-set selection, i.e. choosing set of points such that a model learned over the selected subset is competitive for the remaining data points.We further present a theoretical result characterizing the performance of any selected subset using the geometry of the datapoints.As an active learning algorithm, we choose the subset which is expected to yield best result according to our characterization.Our experiments show that the proposed method significantly outperforms existing approaches in image classification experiments by a large margin.",We approach to the problem of active learning as a core-set selection problem and show that this approach is especially useful in the batch active learning setting which is crucial when training CNNs.The authors provide an algorithm-agnostic active learning algorithm for multi-class classificationThe paper proposes a batch mode active learning algorithm for CNN as a core-set problem which outperforms random sampling and uncertainty sampling.Studies active learning for convolutional neural networks and formulates the active learning problem as core-set selection and presents a novel strategy
244,h-detach: Modifying the LSTM Gradient Towards Better Optimization,"Recurrent neural networks are known for their notorious exploding and vanishing gradient problem.This problem becomes more evident in tasks where the information needed to correctly solve them exist over long time scales, because EVGP prevents important gradient components from being back-propagated adequately over a large number of steps.We introduce a simple stochastic algorithm that is specific to LSTM optimization and targeted towards addressing this problem.Specifically, we show that when the LSTM weights are large, the gradient components through the linear path in the LSTM computational graph get suppressed.Based on the hypothesis that these components carry information about long term dependencies, their suppression can prevent LSTMs from capturing them.Our algorithmfootnote prevents gradients flowing through this path from getting suppressed, thus allowing the LSTM to capture such dependencies better.We show significant improvements over vanilla LSTM gradient based training in terms of convergence speed, robustness to seed and learning rate, and generalization using our modification of LSTM gradient on various benchmark datasets.","A simple algorithm to improve optimization and handling of long term dependencies in LSTMThe paper introduces a simple stochastic algorithm called h-detach that is specific to LSTM optimization and targeted towards addressing this problem.Proposes a simple modification to the training process of the LSTM to facilitate gradient propogation along cell states, or the ""linear temporal path"""
245,Controlling Over-generalization and its Effect on Adversarial Examples Detection and Generation,"Convolutional Neural Networks significantly improve the state-of-the-art for many applications, especially in computer vision.However, CNNs still suffer from a tendency to confidently classify out-distribution samples from unknown classes into pre-defined known classes.Further, they are also vulnerable to adversarial examples.We are relating these two issues through the tendency of CNNs to over-generalize for areas of the input space not covered well by the training set.We show that a CNN augmented with an extra output class can act as a simple yet effective end-to-end model for controlling over-generalization.As an appropriate training set for the extra class, we introduce two resources that are computationally efficient to obtain: a representative natural out-distribution set and interpolated in-distribution samples.To help select a representative natural out-distribution set among available ones, we propose a simple measurement to assess an out-distribution set's fitness.We also demonstrate that training such an augmented CNN with representative out-distribution natural datasets and some interpolated samples allows it to better handle a wide range of unseen out-distribution samples and black-box adversarial examples without training it on any adversaries.Finally, we show that generation of white-box adversarial attacks using our proposed augmented CNN can become harder, as the attack algorithms have to get around the rejection regions when generating actual adversaries.",Properly training CNNs with dustbin class increase their robustness to adversarial attacks and their capacity to deal with out-distribution samples.This paper proposes adding an additional label for detecting OOD samples and adversarial examples in CNN models.The paper proposes an additional class that incorporates natural out-distribution images and interpolated images for adversarial and out-distribution samples in CNNs
246,Data augmentation instead of explicit regularization,"Modern deep artificial neural networks have achieved impressive results through models with very large capacity compared to the number of training examples that control overfitting with the help of different forms of regularization.Regularization can be implicit, as is the case of stochastic gradient descent or parameter sharing in convolutional layers, or explicit.Most common explicit regularization techniques, such as dropout and weight decay, reduce the effective capacity of the model and typically require the use of deeper and wider architectures to compensate for the reduced capacity.Although these techniques have been proven successful in terms of results, they seem to waste capacity.In contrast, data augmentation techniques reduce the generalization error by increasing the number of training examples and without reducing the effective capacity.In this paper we systematically analyze the effect of data augmentation on some popular architectures and conclude that data augmentation alone without any other explicit regularization techniques can achieve the same performance or higher as regularized models, especially when training with fewer examples.","In a deep convolutional neural network trained with sufficient level of data augmentation, optimized by SGD, explicit regularizers (weight decay and dropout) might not provide any additional generalization improvement.This paper proposes data augmentation as an alternative to commonly used regularisation techniques, and shows that for a few reference models/tasks that the same generalization performance can be achived using only data augmentation.This paper presents a systematic study of data augmentation in image classification with deep neural networks, suggesting that data augmentation can replicit some common regularizers like weight decay and dropout."
247,Gedit: Keyboard Gestures for Mobile Text Editing,"Text editing on mobile devices can be a tedious process.To perform various editing operations, a user must repeatedly move his or her fingers between the text input area and the keyboard, making multiple round trips and breaking the flow of typing.In this work, we present Gedit, a system of on-keyboard gestures for convenient mobile text editing.Our design includes a ring gesture and flicks for cursor control, bezel gestures for mode switching, and four gesture shortcuts for copy, paste, cut, and undo.Variations of our gestures exist for one and two hands.We conducted an experiment to compare Gedit with the de facto touch+widget based editing interactions.Our results showed that Gedit’s gestures were easy to learn, 24% and 17% faster than the de facto interactions for one- and two-handed use, respectively, and preferred by participants.","In this work, we present Gedit, a system of on-keyboard gestures for convenient mobile text editing.Reports the design and evaluation of the Gedit interaction techniques.Presents a new set of touch gestures to perform seamless transition between text entry and text editing in mobile devices"
248,Understanding Deep Learning Generalization by Maximum Entropy,"Deep learning achieves remarkable generalization capability with overwhelming number of model parameters.Theoretical understanding of deep learning generalization receives recent attention yet remains not fully explored.This paper attempts to provide an alternative understanding from the perspective of maximum entropy.We first derive two feature conditions that softmax regression strictly apply maximum entropy principle.DNN is then regarded as approximating the feature conditions with multilayer feature learning, and proved to be a recursive solution towards maximum entropy principle.The connection between DNN and maximum entropy well explains why typical designs such as shortcut and regularization improves model generalization, and provides instructions for future model development.",We prove that DNN is a recursively approximated solution to the maximum entropy principle.Presents a derivation which links a DNN to recursive application of maximum entropy model fitting.The paper aims to provide a view of deep learning from the perspective of maximum entropy principle.
249,Investigating CNNs' Learning Representation under label noise,"Deep convolutional neural networks are known to be robust against label noise on extensive datasets.However, at the same time, CNNs are capable of memorizing all labels even if they are random, which means they can memorize corrupted labels.Are CNNs robust or fragile to label noise?Much of researches focusing on such memorization uses class-independent label noise to simulate label corruption, but this setting is simple and unrealistic.In this paper, we investigate the behavior of CNNs under class-dependently simulated label noise, which is generated based on the conceptual distance between classes of a large dataset.Contrary to previous knowledge, we reveal CNNs are more robust to such class-dependent label noise than class-independent label noise.We also demonstrate the networks under class-dependent noise situations learn similar representation to the no noise situation, compared to class-independent noise situations.","Are CNNs robust or fragile to label noise? Practically, robust.The authors challenge the CNNs robustness to label noise using ImageNet 1k tree of WordNet.An analysis of convolutional neural network model performance when class dependent and class independent noise is introducedDemonstrates that CNNs are more robust to class-relevant label noise and argues that real-world noise should be class-relevant"
250,GANSynth: Adversarial Neural Audio Synthesis,"Efficient audio synthesis is an inherently difficult machine learning task, as human perception is sensitive to both global structure and fine-scale waveform coherence.Autoregressive models, such as WaveNet, model local structure at the expense of global latent structure and slow iterative sampling, while Generative Adversarial Networks, have global latent conditioning and efficient parallel sampling, but struggle to generate locally-coherent audio waveforms.Herein, we demonstrate that GANs can in fact generate high-fidelity and locally-coherent audio by modeling log magnitudes and instantaneous frequencies with sufficient frequency resolution in the spectral domain.Through extensive empirical investigations on the NSynth dataset, we demonstrate that GANs are able to outperform strong WaveNet baselines on automated and human evaluation metrics, and efficiently generate audio several orders of magnitude faster than their autoregressive counterparts.","High-quality audio synthesis with GANsProposes an approach that uses GAN framework to generate audio through modeling log magnitudes and instantaneous frequencies with sufficient frequency resolution in the spectral domain. A strategy to generate audio samples from noise with GANs, with changes to the architecture and representation necessary to generate convincing audio that contains an interpretable latent code.Presents a simple idea for better representing audio data so that convolutional models such as generative adversarial networks can be applied"
251,UPS: optimizing Undirected Positive Sparse graph for neural graph filtering,"In this work we propose a novel approach for learning graph representation of the data using gradients obtained via backpropagation.Next we build a neural network architecture compatible with our optimization approach and motivated by graph filtering in the vertex domain.We demonstrate that the learned graph has richer structure than often used nearest neighbors graphs constructed based on features similarity.Our experiments demonstrate that we can improve prediction quality for several convolution on graphs architectures, while others appeared to be insensitive to the input graph.",Graph Optimization with signal filtering in the vertex domain.The paper investigates learning adjacency matrix of a graph with sparsely connected undirected graph with nonnegative edge weights uses a projected sub-gradient descent algorithm.Develops a novel scheme for backpropogating on the adjacency matrix of a neural network graph
252,WAAT: a Workstation AR Authoring Tool for Industry 4.0,"The use of AR in an industrial context could help for the training of new operators.To be able to use an AR guidance system, we need a tool to quickly create a 3D representation of the assembly line and of its AR annotations.This tool should be very easy to use by an operator who is not an AR or VR specialist: typically the manager of the assembly line.This is why we proposed WAAT, a 3D authoring tool allowing user to quickly create 3D models of the workstations, and also test the AR guidance placement.WAAT makes on-site authoring possible, which should really help to have an accurate 3D representation of the assembly line.The verification of AR guidance should also be very useful to make sure everything is visible and doesn't interfere with technical tasks.In addition to these features, our future work will be directed in the deployment of WAAT into a real boiler assembly line to assess the usability of this solution.",This paper describe a 3D authoring tool for providing AR in assembly lines of industry 4.0The paper addresses how AR authoring tools support training of assembly line systems and proposes an approachAn AR guidance system for industrial assembly lines that allows for on-site authoring of AR content.Presents a system that allows factory workers to be trained more efficiently using augmented reality system. 
253,Understand the dynamics of GANs via Primal-Dual Optimization,"Generative adversarial network is one of the best known unsupervised learning techniques these days due to its superior ability to learn data distributions.In spite of its great success in applications, GAN is known to be notoriously hard to train.The tremendous amount of time it takes to run the training algorithm and its sensitivity to hyper-parameter tuning have been haunting researchers in this area.To resolve these issues, we need to first understand how GANs work.Herein, we take a step toward this direction by examining the dynamics of GANs.We relate a large class of GANs including the Wasserstein GANs to max-min optimization problems with the coupling term being linear over the discriminator.By developing new primal-dual optimization tools, we show that, with a proper stepsize choice, the widely used first-order iterative algorithm in training GANs would in fact converge to a stationary solution with a sublinear rate.The same framework also applies to multi-task learning and distributional robust learning problems.We verify our analysis on numerical examples with both synthetic and real data sets.We hope our analysis shed light on future studies on the theoretical properties of relevant machine learning problems.","We show that, with a proper stepsize choice, the widely used first-order iterative algorithm in training GANs would in fact converge to a stationary solution with a sublinear rate.This paper uses GANs and multi-task learning to provide a convergence guarantee for primal-dual algorithms on certain min-max problems.Analyses the learning dynamics of GANs by formulating the problem as a primal-dual optimisation problem by assuming a limited class of models"
254,Consequentialist conditional cooperation in social dilemmas with imperfect information,"Social dilemmas, where mutual cooperation can lead to high payoffs but participants face incentives to cheat, are ubiquitous in multi-agent interaction.We wish to construct agents that cooperate with pure cooperators, avoid exploitation by pure defectors, and incentivize cooperation from the rest.However, often the actions taken by a partner are unobserved or the consequences of individual actions are hard to predict.We show that in a large class of games good strategies can be constructed by conditioning one's behavior solely on outcomes.We call this consequentialist conditional cooperation.We show how to construct such strategies using deep reinforcement learning techniques and demonstrate, both analytically and experimentally, that they are effective in social dilemmas beyond simple matrix games.We also show the limitations of relying purely on consequences and discuss the need for understanding both the consequences of and the intentions behind an action.","We show how to use deep RL to construct agents that can solve social dilemmas beyond matrix games.Learning to play two-player general-sum games with state with imperfect information Specifies a trigger strategy (CCC) and corresponding algorithm, demonstrating convergence to efficient outcomes in social dilemmas without need for agents to observe each other's actions."
255,Nested Dithered Quantization for Communication Reduction in Distributed Training,"In distributed training, the communication cost due to the transmission of gradientsor the parameters of the deep model is a major bottleneck in scaling up the numberof processing nodes.To address this issue, we propose dithered quantization forthe transmission of the stochastic gradients and show that training with DitheredQuantized Stochastic Gradients is similar to the training with unquantizedSGs perturbed by an independent bounded uniform noise, in contrast to the otherquantization methods where the perturbation depends on the gradients and hence,complicating the convergence analysis.We study the convergence of trainingalgorithms using DQSG and the trade off between the number of quantizationlevels and the training time.Next, we observe that there is a correlation among theSGs computed by workers that can be utilized to further reduce the communicationoverhead without any performance loss.Hence, we develop a simple yet effectivequantization scheme, nested dithered quantized SG, that can reduce thecommunication significantly without requiring the workers communicating extrainformation to each other.We prove that although NDQSG requires significantlyless bits, it can achieve the same quantization variance bound as DQSG.Oursimulation results confirm the effectiveness of training using DQSG and NDQSGin reducing the communication bits or the convergence time compared to theexisting methods without sacrificing the accuracy of the trained model.","The paper proposes and analyzes two quantization schemes for communicating Stochastic Gradients in distributed learning which would reduce communication costs compare to the state of the art while maintaining the same accuracy. The authors propose applying dithered quantization to the stochastic gradients computed through the training process, which improves quantization error and achieves superior results compared to baselines, and propose a nested scheme to reduce communication cost.Authors establish a connection between communication reduction in distributed optimization and dithered quantization and develops two new distributed training algorithms where communication overhead is significantly reduced."
256,A Direct Approach to Robust Deep Learning Using Adversarial Networks,"Deep neural networks have been shown to perform well in many classical machine learning problems, especially in image classification tasks.However, researchers have found that neural networks can be easily fooled, and they are surprisingly sensitive to small perturbations imperceptible to humans. Carefully crafted input images can force a well-trained neural network to provide arbitrary outputs. Including adversarial examples during training is a popular defense mechanism against adversarial attacks.In this paper we propose a new defensive mechanism under the generative adversarial network~ framework.We model the adversarial noise using a generative network, trained jointly with a classification discriminative network as a minimax game.We show empirically that our adversarial network approach works well against black box attacks, with performance on par with state-of-art methods such as ensemble adversarial training and adversarial training with projected gradient descent.","Jointly train an adversarial noise generating network with a classification network to provide better robustness to adversarial attacks.A GAN solution for deep models of classification, faced to white and black box attacks, that produces robust models. The paper proposes a defensive mechanism against adversarial attacks using GANs with generated perturbations used as adversarial examples and a discriminator used to distinguish between them"
257,Ensemble Methods as a Defense to Adversarial Perturbations Against Deep Neural Networks,"Deep learning has become the state of the art approach in many machine learning problems such as classification.It has recently been shown that deep learning is highly vulnerable to adversarial perturbations.Taking the camera systems of self-driving cars as an example, small adversarial perturbations can cause the system to make errors in important tasks, such as classifying traffic signs or detecting pedestrians.Hence, in order to use deep learning without safety concerns a proper defense strategy is required.We propose to use ensemble methods as a defense strategy against adversarial perturbations.We find that an attack leading one model to misclassify does not imply the same for other networks performing the same task.This makes ensemble methods an attractive defense strategy against adversarial attacks.We empirically show for the MNIST and the CIFAR-10 data sets that ensemble methods not only improve the accuracy of neural networks on test data but also increase their robustness against adversarial perturbations.","Using ensemble methods as a defense to adversarial perturbations against deep neural networks.This paper proposes to use ensembling as an adversarial defense mechanism.Empirally investigated the robustness of different deep neural entworks ensembles to the two types of attacks, FGSM and BIM, on two popular datasets, MNIST and CIFAR10"
258,Associative Conversation Model: Generating Visual Information from Textual Information,"In this paper, we propose the Associative Conversation Model that generates visual information from textual information and uses it for generating sentences in order to utilize visual information in a dialogue system without image input.In research on Neural Machine Translation, there are studies that generate translated sentences using both images and sentences, and these studies show that visual information improves translation performance.However, it is not possible to use sentence generation algorithms using images for the dialogue systems since many text-based dialogue systems only accept text input.Our approach generates visual information from input text and generates response text using context vector fusing associative visual information and sentence textual information.A comparative experiment between our proposed model and a model without association showed that our proposed model is generating useful sentences by associating visual information related to sentences.Furthermore, analysis experiment of visual association showed that our proposed model generates visual information effective for sentence generation.",Proposal of the sentence generation method based on fusion between textual information and visual information associated with the textual informationThis work describes a deep learning model for dialogue systems that takes advantage of visual information.This paper proposes a novel dataset for grounded dialog and makes a computational observation that it could help to reason about vision even when performing text-based dialog.Proposes to augment traditional text-based sentence generation/dialogue approaches by incorporating visual information by collecting a bunch of data consisting of both text and associated images or video
259,Contextual Recurrent Convolutional Model for Robust Visual Learning,"Feedforward convolutional neural network has achieved a great success in many computer vision tasks.While it validly imitates the hierarchical structure of biological visual system, it still lacks one essential architectural feature: contextual recurrent connections with feedback, which widely exists in biological visual system.In this work, we designed a Contextual Recurrent Convolutional Network with this feature embedded in a standard CNN structure.We found that such feedback connections could enable lower layers to rethink"" about their representations given the top-down contextual information.We carefully studied the components of this network, and showed its robustness and superiority over feedforward baselines in such tasks as noise image classification, partially occluded object recognition and fine-grained image classification.We believed this work could be an important step to help bridge the gap between computer vision models and real biological visual system.","we proposed a novel contextual recurrent convolutional network with robust property of visual learning This paper introduces feedback connection to enhance feature learning through incorporating context information.The paper proposes to add ""recurrent"" connections inside a convolution network with gating mechanism."
260,Bayesian Uncertainty Estimation for Batch Normalized Deep Networks,"Deep neural networks have led to a series of breakthroughs, dramatically improving the state-of-the-art in many domains.The techniques driving these advances, however, lack a formal method to account for model uncertainty.While the Bayesian approach to learning provides a solid theoretical framework to handle uncertainty, inference in Bayesian-inspired deep neural networks is difficult.In this paper, we provide a practical approach to Bayesian learning that relies on a regularization technique found in nearly every modern network, batch normalization.We show that training a deep network using batch normalization is equivalent to approximate inference in Bayesian models, and we demonstrate how this finding allows us to make useful estimates of the model uncertainty.Using our approach, it is possible to make meaningful uncertainty estimates using conventional architectures without modifying the network or the training procedure.Our approach is thoroughly validated in a series of empirical experiments on different tasks and using various measures, showing it to outperform baselines on a majority of datasets with strong statistical significance.","We show that training a deep network using batch normalization is equivalent to approximate inference in Bayesian models, and we demonstrate how this finding allows us to make useful estimates of the model uncertainty in conventional networks.This paper proposes using batch normalisation at test time to get the predictive uncertainty, and shows Monte Carlo prediction at test time using batch norm is better than dropout.Proposes that the regularization procedure called batch normalization can be understood as performing approximate Bayesian inference, which performs similarly to MC dropout in terms of the estimates of uncertainty that it produces."
261,Combining Global Sparse Gradients with Local Gradients,"Data-parallel neural network training is network-intensive, so gradient dropping was designed to exchange only large gradients. However, gradient dropping has been shown to slow convergence. We propose to improve convergence by having each node combine its locally computed gradient with the sparse global gradient exchanged over the network.We empirically confirm with machine translation tasks that gradient dropping with local gradients approaches convergence 48% faster than non-compressed multi-node training and 28% faster compared to vanilla gradient dropping.We also show that gradient dropping with a local gradient update does not reduce the model's final quality.","We improve gradient dropping (a technique of only exchanging large gradients on distributed training) by incorporating local gradients while doing a parameter update to reduce quality loss and further improve the training time.This paper proposes a 3 modes for combining local and global gradients to better use more computing nodesLooks at the problem of reducing the communication requirement for implementing the distributed optimiztion techniques, particularly SGD"
262,Exploration using Distributional RL and UCB," We establish the relation between Distributional RL and the Upper Confidence Bound approach to exploration. In this paper we show that the density of the Q function estimated by Distributional RL can be successfully used for the estimation of UCB.This approach does not require counting and, therefore, generalizes well to the Deep RL.We also point to the asymmetry of the empirical densities estimated by the Distributional RL algorithms like QR-DQN.This observation leads to the reexamination of the variance's performance in the UCB type approach to exploration.We introduce truncated variance as an alternative estimator of the UCB and a novel algorithm based on it.We empirically show that newly introduced algorithm achieves better performance in multi-armed bandits setting.Finally, we extend this approach to high-dimensional setting and test it on the Atari 2600 games.New approach achieves better performance compared to QR-DQN in 26 of games, 13 ties out of 49 games.","Exploration using Distributional RL and truncagted variance.Presents an RL method to manage exploration-explotation trade-offs via UCB techniques.A method to use the distribution learned by Quantile Regression DQN for exploration, in place of the usual epsilon-greedy strategy.Proposes new algorithsms (QUCB and QUCB+) to handle the exploration tradeoff in Multi-Armed Bendits and more generally in Reinforcement Learning"
263,Shaping representations through communication,"Good representations facilitate transfer learning and few-shot learning.Motivated by theories of language and communication that explain why communities with large number of speakers have, on average, simpler languages with more regularity, we cast the representation learning problem in terms of learning to communicate.Our starting point sees traditional autoencoders as a single encoder with a fixed decoder partner that must learn to communicate.Generalizing from there, we introduce community-based autoencoders in which multiple encoders and decoders collectively learn representations by being randomly paired up on successive training iterations.Our experiments show that increasing community sizes reduce idiosyncrasies in the learned codes, resulting in more invariant representations with increased reusability and structure.","Motivated by theories of language and communication, we introduce community-based autoencoders, in which multiple encoders and decoders collectively learn structured and reusable representations.The authors tackle the problem of representation learning, aim to build reusable and structured represenation, argue co-adaptation between encoder and decoder in traditional AE yields poor representation, and introduce community based auto-encoders.The paper presents a community based autoencoder framework to address co-adaptation of encoders and decoders and aims at constructing better representations."
264,One-Shot High-Fidelity Imitation: Training Large-Scale Deep Nets with RL,"Humans are experts at high-fidelity imitation -- closely mimicking a demonstration, often in one attempt.Humans use this ability to quickly solve a task instance, and to bootstrap learning of new tasks.Achieving these abilities in autonomous agents is an open problem.In this paper, we introduce an off-policy RL algorithm to narrow this gap.MetaMimic can learn both policies for high-fidelity one-shot imitation of diverse novel skills, and policies that enable the agent to solve tasks more efficiently than the demonstrators.MetaMimic relies on the principle of storing all experiences in a memory and replaying these to learn massive deep neural network policies by off-policy RL.This paper introduces, to the best of our knowledge, the largest existing neural networks for deep RL and shows that larger networks with normalization are needed to achieve one-shot high-fidelity imitation on a challenging manipulation task.The results also show that both types of policy can be learned from vision, in spite of the task rewards being sparse, and without access to demonstrator actions.","We present MetaMimic, an algorithm that takes as input a demonstration dataset and outputs (i) a one-shot high-fidelity imitation policy (ii) an unconditional task policy.The paper looks at the problem of one-shot imitation with high accuracy of imitation, extending DDPGfD to use only state trajectories.This paper proposes an approach for one-shot imitation with high accuracy, and addresses the common problem of exploration in imitation learning.Presents an RL method for learning from video demonstration without access to expert actions"
265,Mode Normalization,"Normalization methods are a central building block in the deep learning toolbox.They accelerate and stabilize training, while decreasing the dependence on manually tuned learning rate schedules.When learning from multi-modal distributions, the effectiveness of batch normalization, arguably the most prominent normalization method, is reduced.As a remedy, we propose a more flexible approach: by extending the normalization to more than a single mean and variance, we detect modes of data on-the-fly, jointly normalizing samples that share common features.We demonstrate that our method outperforms BN and other widely used normalization techniques in several experiments, including single and multi-task datasets.",We present a novel normalization method for deep neural networks that is robust to multi-modalities in intermediate feature distributions.Normalization method that learns multi-modal distribution in the feature spaceProposes a generalization of Batch Normalization under the assumption that the statistics of the unit activations over the batches and over the spatial dimensions is not unimodal
266,Multilingual Neural Machine Translation with Knowledge Distillation,"Multilingual machine translation, which translates multiple languages with a single model, has attracted much attention due to its efficiency of offline training and online serving.However, traditional multilingual translation usually yields inferior accuracy compared with the counterpart using individual models for each language pair, due to language diversity and model capacity limitations.In this paper, we propose a distillation-based approach to boost the accuracy of multilingual machine translation.Specifically, individual models are first trained and regarded as teachers, and then the multilingual model is trained to fit the training data and match the outputs of individual models simultaneously through knowledge distillation.Experiments on IWSLT, WMT and Ted talk translation datasets demonstrate the effectiveness of our method.Particularly, we show that one model is enough to handle multiple languages, with comparable or even better accuracy than individual models.",We proposed a knowledge distillation based method to boost the accuracy of multilingual neural machine translation.A many-to-one multilingual neural machine translation model that first training separate models for each language pair then performs distillation.The paper aims at training a machine translation model by augmenting the standard cross-entropy loss with a distillation component based on individual (single-language-pair) teacher models.
267,Investigating Human Priors for Playing Video Games,"What makes humans so good at solving seemingly complex video games? Unlike computers, humans bring in a great deal of prior knowledge about the world, enabling efficient decision making.This paper investigates the role of human priors for solving video games.Given a sample game, we conduct a series of ablation studies to quantify the importance of various priors.We do this by modifying the video game environment to systematically mask different types of visual information that could be used by humans as priors.We find that removal of some prior knowledge causes a drastic degradation in the speed with which human players solve the game, e.g. from 2 minutes to over 20 minutes.Furthermore, our results indicate that general priors, such as the importance of objects and visual consistency, are critical for efficient game-play.","We investigate the various kinds of prior knowledge that help human learning and find that general priors about objects play the most critical role in guiding human gameplay.The authors study by experiment, what aspects of human priors are the important for reinforcement learning in video games.The authors present a study of priors employed by humans in playing video games and demonstrates the existence of a taxonomy of features that affect the ability to complete tasks in the game to varying degrees."
268,Open Loop Hyperparameter Optimization and Determinantal Point Processes,"Driven by the need for parallelizable hyperparameter optimization methods, this paper studies search methods: sequences that are predetermined and can be generated before a single configuration is evaluated.Examples include grid search, uniform random search, low discrepancy sequences, and other sampling distributions.In particular, we propose the use of-determinantal point processes in hyperparameter optimization via random search.Compared to conventional uniform random search where hyperparameter settings are sampled independently, a-DPP promotes diversity. We describe an approach that transforms hyperparameter search spaces for efficient use with a-DPP.In addition, we introduce a novel Metropolis-Hastings algorithm which can sample from-DPPs defined over spaces with a mixture of discrete and continuous dimensions.Our experiments show significant benefits over uniform random search in realistic scenarios with a limited budget for training supervised learners, whether in serial or parallel.","Driven by the need for parallelizable, open-loop hyperparameter optimization methods, we propose the use of-determinantal point processes in hyperparameter optimization via random search.Proposes using the k-DPP to select candidate points in hyperparameter searches.The authors propose k-DPP as an open loop method for hyperparameter optimization and provide its empirical study and comparison with other methods. Considers non-sequential and uninformed hyperparameter search using determinantal point processes, which are probability distributions over subsets of a ground set with the property that subsets with more 'diverse' elements haev higher probability"
269, Explicit Induction Bias for Transfer Learning with Convolutional Networks,"In inductive transfer learning, fine-tuning pre-trained convolutional networks substantially outperforms training from scratch.When using fine-tuning, the underlying assumption is that the pre-trained model extracts generic features, which are at least partially relevant for solving the target task, but would be difficult to extract from the limited amount of data available on the target task.However, besides the initialization with the pre-trained model and the early stopping, there is no mechanism in fine-tuning for retaining the features learned on the source task.In this paper, we investigate several regularization schemes that explicitly promote the similarity of the final solution with the initial model.We eventually recommend a simple penalty using the pre-trained model as a reference, and we show that this approach behaves much better than the standard scheme using weight decay on a partially frozen network.","In inductive transfer learning, fine-tuning pre-trained convolutional networks substantially outperforms training from scratch.Addresses the problem of transfer learning in deep networks and proposes to have a regularization term that penalizes divergence from initialization.Proposes an analysis on different adaptive regularization techniques for deep transfer learning, specifically focusing on the use of an L-SP condition"
270,Neural Networks with Block Diagonal Inner Product Layers,"Artificial neural networks have opened up a world of possibilities in data science and artificial intelligence, but neural networks are cumbersome tools that grow with the complexity of the learning problem.We make contributions to this issue by considering a modified version of the fully connected layer we call a block diagonal inner product layer.These modified layers have weight matrices that are block diagonal, turning a single fully connected layer into a set of densely connected neuron groups.This idea is a natural extension of group, or depthwise separable, convolutional layers applied to the fully connected layers.Block diagonal inner product layers can be achieved by either initializing a purely block diagonal weight matrix or by iteratively pruning off diagonal block entries.This method condenses network storage and speeds up the run time without significant adverse effect on the testing accuracy, thus offering a new approach to improve network computation efficiency.","We look at neural networks with block diagonal inner product layers for efficiency.This paper proposes making the inner layers in a neural network be block diagonal, and discusses that block diagonal matrices are more efficient than pruning and block diagonal layers lead to more efficient networks.Replacing fully connected layers with block-diagonal fully connected layers"
271,Spectral Normalization for Generative Adversarial Networks,"One of the challenges in the study of generative adversarial networks is the instability of its training.In this paper, we propose a novel weight normalization technique called spectral normalization to stabilize the training of the discriminator.Our new normalization technique is computationally light and easy to incorporate into existing implementations.We tested the efficacy of spectral normalization on CIFAR10, STL-10, and ILSVRC2012 dataset, and we experimentally confirmed that spectrally normalized GANs is capable of generating images of better or equal quality relative to the previous training stabilization techniques.","We propose a novel weight normalization technique called spectral normalization to stabilize the training of the discriminator of GANs.This paper uses spectral regularization to normalize GAN objectives, and the ensuing GAN, called SN-GAN, essentially ensures the Lipschitz property of the discriminator.This paper proposes""spectral normalization"", moving a nice step forward in improving the training of GANs."
272,Composing Complex Skills by Learning Transition Policies,"Humans acquire complex skills by exploiting previously learned skills and making transitions between them.To empower machines with this ability, we propose a method that can learn transition policies which effectively connect primitive skills to perform sequential tasks without handcrafted rewards.To efficiently train our transition policies, we introduce proximity predictors which induce rewards gauging proximity to suitable initial states for the next skill.The proposed method is evaluated on a set of complex continuous control tasks in bipedal locomotion and robotic arm manipulation which traditional policy gradient methods struggle at.We demonstrate that transition policies enable us to effectively compose complex skills with existing primitive skills.The proposed induced rewards computed using the proximity predictor further improve training efficiency by providing more dense information than the sparse rewards from the environments.We make our environments, primitive skills, and code public for further research at https://youngwoon.github.io/transition .",Transition policies enable agents to compose complex skills by smoothly connecting previously acquired primitive skills.Proposes a scheme for transitioning to favorable strating states for executing given options in continuous domains. This uses two learning processes carried out simultaneously.Presents a method for learning policies for transitioning from one task to another with the goal of completing complex tasks using state proximity estimator to reward for transition policy.Proposes a new training scheme with a learned auxiliary reward function to optimise transition policies that connect the ending state of a previous macro action/option with good initiation states of the following macro action/option
273,The Expressive Power of Gated Recurrent Units as a Continuous Dynamical System,"Gated recurrent units were inspired by the common gated recurrent unit, long short-term memory, as a means of capturing temporal structure with less complex memory unit architecture.Despite their incredible success in tasks such as natural and artificial language processing, speech, video, and polyphonic music, very little is understood about the specific dynamic features representable in a GRU network.As a result, it is difficult to know a priori how successful a GRU-RNN will perform on a given data set.In this paper, we develop a new theoretical framework to analyze one and two dimensional GRUs as a continuous dynamical system, and classify the dynamical features obtainable with such system.We found rich repertoire that includes stable limit cycles over time, multi-stable state transitions with various topologies, and homoclinic orbits.In addition, we show that any finite dimensional GRU cannot precisely replicate the dynamics of a ring attractor, or more generally, any continuous attractor, and is limited to finitely many isolated fixed points in theory.These findings were then experimentally verified in two dimensions by means of time series prediction.","We classify the the dynamical features one and two GRU cells can and cannot capture in continuous time, and verify our findings experimentally with k-step time series prediction. The authors analyse GRUs with hidden sizes of one and two as continuous-time dynamical systems, claiming that the expressive power of the hidden state representation can provide prior knowledge on how well a GRU will perform on a given datasetThis paper analyzes GRUs from a dynamical systems perspective, and shows that 2d GRUs can be trained to adopt a variety of fixed points and can approximate line attractors, but cannot mimic a ring attractor.Converts GRU equations into continuous time and uses theory and experiemnts to study 1- and 2-dimensional GRU networks and showcase every variety of dynamical topology available in these systems"
274,Multi-Scale Stacked Hourglass Network for Human Pose Estimation,"Stacked hourglass network has become an important model for Human pose estimation.The estimation of human body posture depends on the global information of the keypoints type and the local information of the keypoints location.The consistent processing of inputs and constraints makes it difficult to form differentiated and determined collaboration mechanisms for each stacked hourglass network.In this paper, we propose a Multi-Scale Stacked Hourglass network to high-light the differentiation capabilities of each Hourglass network for human pose estimation. The pre-processing network forms feature maps of different scales,and dispatch them to various locations of the stack hourglass network, where the small-scale features reach the front of stacked hourglass network, and large-scale features reach the rear of stacked hourglass network. And a new loss function is proposed for multi-scale stacked hourglass network. Different keypoints have different weight coefficients of loss function at different scales, and the keypoints weight coefficients are dynamically adjusted from the top-level hourglass network to the bottom-level hourglass network. Experimental results show that the pro-posed method is competitive with respect to the comparison algorithm on MPII and LSP datasets.","Differentiated inputs cause functional differentiation of the network, and the interaction of loss functions between networks can affect the optimization process.A modification to the original hourglass network for single pose estimation that yields improvements over the original baseline.Authors extend a stacked hourglass network with inception-resnet-A modules and propose a multi-scale approach for human pose estimation in still RGB images."
275,UNSUPERVISED SENTENCE EMBEDDING USING DOCUMENT STRUCTURE-BASED CONTEXT,"We present a new unsupervised method for learning general-purpose sentence embeddings.Unlike existing methods which rely on local contexts, such as wordsinside the sentence or immediately neighboring sentences, our method selects, foreach target sentence, influential sentences in the entire document based on a documentstructure.We identify a dependency structure of sentences using metadataor text styles.Furthermore, we propose a novel out-of-vocabulary word handlingtechnique to model many domain-specific terms, which were mostly discarded byexisting sentence embedding methods.We validate our model on several tasksshowing 30% precision improvement in coreference resolution in a technical domain,and 7.5% accuracy increase in paraphrase detection compared to baselines.","To train a sentence embedding using technical documents, our approach considers document structure to find broader context and handle out-of-vocabulary words.Presents ideas for improving sentence embedding by drawing from more context.Learning sentence representations with sentences dependencies informationExtends the idea of forming an unsupervised representation of sentences used in the SkipThough approach by using a broader set of evidence for forming the representation of a sentence"
276,Visualizing the Loss Landscape of Neural Nets,"Neural network training relies on our ability to find ""good"" minimizers of highly non-convex loss functions.It is well known that certain network architecture designs produce loss functions that train easier, and well-chosen training parameters produce minimizers that generalize better.However, the reasons for these differences, and their effect on the underlying loss landscape, is not well understood.In this paper, we explore the structure of neural loss functions, and the effect of loss landscapes on generalization, using a range of visualization methods.First, we introduce a simple ""filter normalization"" method that helps us visualize loss function curvature, and make meaningful side-by-side comparisons between loss functions.Then, using a variety of visualizations, we explore how network architecture effects the loss landscape, and how training parameters affect the shape of minimizers.","We explore the structure of neural loss functions, and the effect of loss landscapes on generalization, using a range of visualization methods.This paper proposes a method to visualize the loss function of a NN and provides insights on the trainability and generalization of NNs.Investigates the non-convexity of the loss surface and optimization paths."
277,Multi-way Encoding for Robustness to Adversarial Attacks,"Deep models are state-of-the-art for many computer vision tasks including image classification and object detection.However, it has been shown that deep models are vulnerable to adversarial examples.We highlight how one-hot encoding directly contributes to this vulnerability and propose breaking away from this widely-used, but highly-vulnerable mapping.We demonstrate that by leveraging a different output encoding, multi-way encoding, we can make models more robust.Our approach makes it more difficult for adversaries to find useful gradients for generating adversarial attacks.We present state-of-the-art robustness results for black-box, white-box attacks, and achieve higher clean accuracy on four benchmark datasets: MNIST, CIFAR-10, CIFAR-100, and SVHN when combined with adversarial training.The strength of our approach is also presented in the form of an attack for model watermarking, raising challenges in detecting stolen models.","We demonstrate that by leveraging a multi-way output encoding, rather than the widely used one-hot encoding, we can make deep models more robust to adversarial attacks.This paper proposes replacing the final cross-entropy layer trained on one-hot labels in classifiers by encoding each label as a high-dimensional vector and training the classifier to minimize L2 distance from the encoding of the correct class.Authors propose new method against adversarial attacks that shows significant amount of gains compared to baselines"
278,Non-Autoregressive Neural Machine Translation,"Existing approaches to neural machine translation condition each output word on previously generated outputs.We introduce a model that avoids this autoregressive property and produces its outputs in parallel, allowing an order of magnitude lower latency during inference.Through knowledge distillation, the use of input token fertilities as a latent variable, and policy gradient fine-tuning, we achieve this at a cost of as little as 2.0 BLEU points relative to the autoregressive Transformer network used as a teacher.We demonstrate substantial cumulative improvements associated with each of the three aspects of our training strategy, and validate our approach on IWSLT 2016 English–German and two WMT language pairs.By sampling fertilities in parallel at inference time, our non-autoregressive model achieves near-state-of-the-art performance of 29.8 BLEU on WMT 2016 English–Romanian.","We introduce the first NMT model with fully parallel decoding, reducing inference latency by 10x.This work proposes non-autoregressive decoder for the encoder-decoder framework in which the decision of generating a word does not depends on the prior decision of generated wordsThis paper describes an approach to decode non-autoregressively for neural machine translation with the possibility of more parallel decoding which can result in a significant speed-up.Proposes the introduction of a set of latent variables to represent the fertility of each source word to make the target sentence generation non-autoregressive"
279,Certified Defenses against Adversarial Examples ,"While neural networks have achieved high accuracy on standard image classification benchmarks, their accuracy drops to nearly zero in the presence of small adversarial perturbations to test inputs.Defenses based on regularization and adversarial training have been proposed, but often followed by new, stronger attacks that defeat these defenses.Can we somehow end this arms race?In this work, we study this problem for neural networks with one hidden layer.We first propose a method based on a semidefinite relaxation that outputs a certificate that for a given network and test input, no attack can force the error to exceed a certain value.Second, as this certificate is differentiable, we jointly optimize it with the network parameters, providing an adaptive regularizer that encourages robustness against all attacks.On MNIST, our approach produces a network and a certificate that no that perturbs each pixel by at most can cause more than test error.","We demonstrate a certifiable, trainable, and scalable method for defending against adversarial examples.Proposes a new defense against security attacks on neural networks with the atack model that outputs a security certificate on the algorithm.Derives an upper bound on adversarial perturbation for neural networks with one hidden layer"
280,Information Regularized Neural Networks,"We formulate an information-based optimization problem for supervised classification.For invertible neural networks, the control of these information terms is passed down to the latent features and parameter matrix in the last fully connected layer, given that mutual information is invariant under invertible map. We propose an objective function and prove that it solves the optimization problem.Our framework allows us to learn latent features in an more interpretable form while improving the classification performance.We perform extensive quantitative and qualitative experiments in comparison with the existing state-of-the-art classification models.","we propose a regularizer that improves the classification performance of neural networksthe authors propose to train a model from a point of maximizing mutual information between the predictions and the true outputs, with a regularization term that minimizes irrelevant information while learning.Proposes to decompose the parameters into an invertible feature map F and a linear transformation w in the last layer to maximize mutual information I(Y, hat) while constraining irrelevant information"
281,Improving latent variable descriptiveness by modelling rather than ad-hoc factors,"Powerful generative models, particularly in Natural Language Modelling, are commonly trained by maximizing a variational lower bound on the data log likelihood.These models often suffer from poor use of their latent variable, with ad-hoc annealing factors used to encourage retention of information in the latent variable.We discuss an alternative and general approach to latent variable modelling, based on an objective that encourages a perfect reconstruction by tying a stochastic autoencoder with a variational autoencoder.This ensures by design that the latent variable captures information about the observations, whilst retaining the ability to generate well.Interestingly, although our model is fundamentally different to a VAE, the lower bound attained is identical to the standard VAE bound but with the addition of a simple pre-factor; thus, providing a formal interpretation of the commonly used, ad-hoc pre-factors in training VAEs.","This paper introduces a novel generative modelling framework that avoids latent-variable collapse and clarifies the use of certain ad-hoc factors in training Variational Autoencoders.The paper proposes to resolve the issue about a variational auto-encoder ignoring the latent variables. This paper proposes adding a stochastic autoencoder to the original VAE model to address the problem that the LSTM decoder of a language model might be too strong to ignore the latent variable's information."", 'This paper presents AutoGen, which combines a generative variational autoencoder with a high-fidelity reconstruction model based on autoencoder to better utiliza latent representation"
282,Learning to Separate Domains in Generalized Zero-Shot and Open Set Learning: a probabilistic perspective,"This paper studies the problem of domain division which aims to segment instances drawn from different probabilistic distributions.This problem exists in many previous recognition tasks, such as Open Set Learning and Generalized Zero-Shot Learning, where the testing instances come from either seen or unseen/novel classes with different probabilistic distributions.Previous works only calibrate the confident prediction of classifiers of seen classes) or taking unseen classes as outliers Socher et al..In contrast, this paper proposes a probabilistic way of directly estimating and fine-tuning the decision boundary between seen and unseen classes.In particular, we propose a domain division algorithm to split the testing instances into known, unknown and uncertain domains, and then conduct recognition tasks in each domain.Two statistical tools, namely, bootstrapping and KolmogorovSmirnov Test, for the first time, are introduced to uncover and fine-tune the decision boundary of each domain.Critically, the uncertain domain is newly introduced in our framework to adopt those instances whose domain labels cannot be predicted confidently.Extensive experiments demonstrate that our approach achieved the state-of-the-art performance on OSL and G-ZSL benchmarks."," This paper studies the problem of domain division by segmenting instances drawn from different probabilistic distributions. This paper deals with the problem of novelty recognition in open set learning and generalized zero-shot learning and proposes a possible solutionAn approach to domain separation based on bootstrapping to identify similarity cutoff thresholds for known classes, followed by a Kolmogorov-Smirnoff test to refine the bootstrapped in-distribution zones.Proposes to introduce a new domain, the uncertain domain, to better handle the division between seen/unseen domains in open-set and generalized zero-shot learning"
283,"Stochastic gradient descent performs variational inference, converges to limit cycles for deep networks","Stochastic gradient descent is widely believed to perform implicit regularization when used to train deep neural networks, but the precise manner in which this occurs has thus far been elusive.We prove that SGD minimizes an average potential over the posterior distribution of weights along with an entropic regularization term.This potential is however not the original loss function in general.So SGD does perform variational inference, but for a different loss than the one used to compute the gradients.Even more surprisingly, SGD does not even converge in the classical sense: we show that the most likely trajectories of SGD for deep networks do not behave like Brownian motion around critical points.Instead, they resemble closed loops with deterministic components.We prove that such out-of-equilibrium behavior is a consequence of highly non-isotropic gradient noise in SGD; the covariance matrix of mini-batch gradients for deep networks has a rank as small as 1% of its dimension.We provide extensive empirical validation of these claims, proven in the appendix.","SGD implicitly performs variational inference; gradient noise is highly non-isotropic, so SGD does not even converge to critical points of the original lossThis paper provides a variational analysis of SGD as a non-equilibrium process.This paper discusses the regularized objective function minimized by standard SGD in the context of neural nets, and provide a variational inference perspective using the Fokker-Planck equation.Develops a theory to study the impact of stocastic gradient noise for SGD, especially for deep neural network models"
284,Zero-Shot Visual Imitation,"The current dominant paradigm for imitation learning relies on strong supervision of expert actions to learn both 'what' and 'how' to imitate.We pursue an alternative paradigm wherein an agent first explores the world without any expert supervision and then distills its experience into a goal-conditioned skill policy with a novel forward consistency loss.In our framework, the role of the expert is only to communicate the goals during inference.The learned policy is then employed to mimic the expert after seeing just a sequence of images demonstrating the desired task.Our method is 'zero-shot' in the sense that the agent never has access to expert actions during training or for the task demonstration at inference.We evaluate our zero-shot imitator in two real-world settings: complex rope manipulation with a Baxter robot and navigation in previously unseen office environments with a TurtleBot.Through further experiments in VizDoom simulation, we provide evidence that better mechanisms for exploration lead to learning a more capable policy which in turn improves end task performance.Videos, models, and more details are available at https://pathak22.github.io/zeroshot-imitation/.","Agents can learn to imitate solely visual demonstrations (without actions) at test time after learning from their own experience without any form of supervision at training time.This paper proposes and approach for zero-shot visual learning by learning parametric skill functions.A paper about imitation of a task presented just during inference, where learning is performed in a self-supervised manner and during training the agent explores related but different tasks.Proposes a method for sidestepping the issue of expensive expert demonstration by using the random exploration of an agent to learn generalizable skills which can be applied without specific pretraining"
285,Comparison of Paragram and GloVe Results for Similarity Benchmarks,"Distributional Semantics Models derive word space from linguistic itemsin context.Meaning is obtained by defining a distance measure between vectorscorresponding to lexical entities.Such vectors present several problems.Thiswork concentrates on quality of word embeddings, improvement of word embeddingvectors, applicability of a novel similarity metric used ‘on top’ of theword embeddings.In this paper we provide comparison between two methodsfor post process improvements to the baseline DSM vectors.The counter-fittingmethod which enforces antonymy and synonymy constraints into the Paragramvector space representations recently showed improvement in the vectors’ capabilityfor judging semantic similarity.The second method is our novel RESMmethod applied to GloVe baseline vectors.By applying the hubness reductionmethod, implementing relational knowledge into the model by retrofitting synonymsand providing a new ranking similarity definition RESM that gives maximumweight to the top vector component values we equal the results for the ESLand TOEFL sets in comparison with our calculations using the Paragram and Paragram+ Counter-fitting methods.For SIMLEX-999 gold standard since we cannotuse the RESM the results using GloVe and PPDB are significantly worse comparedto Paragram.Apparently, counter-fitting corrects hubness.The Paragramor our cosine retrofitting method are state-of-the-art results for the SIMLEX-999gold standard.They are 0.2 better for SIMLEX-999 than word2vec with sensede-conflation.Apparently relational knowledge and counter-fitting is more importantfor judging semantic similarity than sense determination for words.It is tobe mentioned, though that Paragram hyperparameters are fitted to SIMLEX-999results.The lesson is that many corrections to word embeddings are necessaryand methods with more parameters and hyperparameters perform better.",Paper provides a description of a procedure to enhance word vector space model with an evaluation of Paragram and GloVe models for Similarity Benchmarks.This paper suggests a new algorithm that adjusts GloVe word vectors and then uses a non-Euclidean similarity function between them.The authors present observations on the weaknesses of the existing vector space models and list a 6-step approach for refining existing word vectors
286,Alternating Multi-bit Quantization for Recurrent Neural Networks,"Recurrent neural networks have achieved excellent performance in many applications.However, on portable devices with limited resources, the models are often too large to deploy.For applications on the server with large scale concurrent requests, the latency during inference can also be very critical for costly computing resources.In this work, we address these problems by quantizing the network, both weights and activations, into multiple binary codes. We formulate the quantization as an optimization problem.Under the key observation that once the quantization coefficients are fixed the binary codes can be derived efficiently by binary search tree, alternating minimization is then applied. We test the quantization for two well-known RNNs, i.e., long short term memory and gated recurrent unit, on the language models.Compared with the full-precision counter part, by 2-bit quantization we can achieve ~16x memory saving and ~6x real inference acceleration on CPUs, with only a reasonable loss in the accuracy.By 3-bit quantization, we can achieve almost no loss in the accuracy or even surpass the original model, with ~10.5x memory saving and ~3x real inference acceleration.Both results beat the exiting quantization works with large margins. We extend our alternating quantization to image classification tasks.In both RNNs and feedforward neural networks, the method also achieves excellent performance.","We propose a new quantization method and apply it to quantize RNNs for both compression and accelerationThis paper proposes a multi-bit quantization method for recurrent neural networks.A technique for quantizing neural network weight matrices, and an alternating optimization procedure to estimate the set of k binary vectors and coefficients that best represent the original vector."
287,A SINGLE SHOT PCA-DRIVEN ANALYSIS OF NETWORK STRUCTURE TO REMOVE REDUNDANCY,"Deep learning models have outperformed traditional methods in many fields suchas natural language processing and computer vision.However, despite theirtremendous success, the methods of designing optimal Convolutional Neural Networks are still based on heuristics or grid search.The resulting networksobtained using these techniques are often overparametrized with huge computationaland memory requirements.This paper focuses on a structured, explainableapproach towards optimal model design that maximizes accuracy while keepingcomputational costs tractable.We propose a single-shot analysis of a trained CNNthat uses Principal Component Analysis to determine the number of filtersthat are doing significant transformations per layer, without the need for retraining.It can be interpreted as identifying the dimensionality of the hypothesis spaceunder consideration.The proposed technique also helps estimate an optimal numberof layers by looking at the expansion of dimensions as the model gets deeper.This analysis can be used to design an optimal structure of a given network ona dataset, or help to adapt a predesigned network on a new dataset.We demonstratethese techniques by optimizing VGG and AlexNet networks on CIFAR-10,CIFAR-100 and ImageNet datasets.","We present a single shot analysis of a trained neural network to remove redundancy and identify optimal network structureThis paper proposes a set of heuristics for identifying a good neural network architecture, based on PCA of unit activations over the datasetThis paper presents a framework for optimising neural networks architectures through the identification of redundant filters across layers"
288,Security Analysis of Deep Neural Networks Operating in the Presence of Cache Side-Channel Attacks,"Recent work has introduced attacks that extract the architecture information of deep neural networks, as this knowledge enhances an adversary’s capability to conduct attacks on black-box networks.This paper presents the first in-depth security analysis of DNN fingerprinting attacks that exploit cache side-channels. First, we define the threat model for these attacks: our adversary does not need the ability to query the victim model; instead, she runs a co-located process on the host machine victim ’s deep learning system is running and passively monitors the accesses of the target functions in the shared framework. Second, we introduce DeepRecon, an attack that reconstructs the architecture of the victim network by using the internal information extracted via Flush+Reload, a cache side-channel technique.Once the attacker observes function invocations that map directly to architecture attributes of the victim network, the attacker can reconstruct the victim’s entire network architecture. In our evaluation, we demonstrate that an attacker can accurately reconstruct two complex networks having only observed one forward propagation.Based on the extracted architecture attributes, we also demonstrate that an attacker can build a meta-model that accurately fingerprints the architecture and family of the pre-trained model in a transfer learning setting.From this meta-model, we evaluate the importance of the observed attributes in the fingerprinting process.Third, we propose and evaluate new framework-level defense techniques that obfuscate our attacker’s observations.Our empirical security analysis represents a step toward understanding the DNNs’ vulnerability to cache side-channel attacks.","We conduct the first in-depth security analysis of DNN fingerprinting attacks that exploit cache side-channels, which represents a step toward understanding the DNN’s vulnerability to side-channel attacks.This paper considers the problem of fingerprinting neural network architectures using cache side channels, and discusses security-through-obscurity defenses.This paper performs cache side-channel attacks to extract attributes of a victim model and infer its architecture, as well as show they can achieve a nearly perfect classification accuracy."
289,Complement Objective Training,"Learning with a primary objective, such as softmax cross entropy for classification and sequence generation, has been the norm for training deep neural networks for years.Although being a widely-adopted approach, using cross entropy as the primary objective exploits mostly the information from the ground-truth class for maximizing data likelihood, and largely ignores information from the complement classes.We argue that, in addition to the primary objective, training also using a complement objective that leverages information from the complement classes can be effective in improving model performance.This motivates us to study a new training paradigm that maximizes the likelihood of the ground-truth class while neutralizing the probabilities of the complement classes.We conduct extensive experiments on multiple tasks ranging from computer vision to natural language understanding.The experimental results confirm that, compared to the conventional training with just one primary objective, training also with the complement objective further improves the performance of the state-of-the-art models across all tasks.In addition to the accuracy improvement, we also show that models trained with both primary and complement objectives are more robust to single-step adversarial attacks.","We propose Complement Objective Training (COT), a new training paradigm that optimizes both the primary and complement objectives for effectively learning the parameters of neural networks.Considers augmenting the cross-entropy objective with ""complement"" objective maximization, which aims at neutralizing the predicted probabilities of classes other than the ground truth labels.The authors propose a secondary objective for softmax minimization based on evaluating the information gathered from the incorrect classes, leading to a new training approach.Deals with the training of neural networks for classification or sequence generation tasks using across-entropy loss"
290,Inhibited Softmax for Uncertainty Estimation in Neural Networks,We present a new method for uncertainty estimation and out-of-distribution detection in neural networks with softmax output.We extend softmax layer with an additional constant input.The corresponding additional output is able to represent the uncertainty of the network.The proposed method requires neither additional parameters nor multiple forward passes nor input preprocessing nor out-of-distribution datasets.We show that our method performs comparably to more computationally expensive methods and outperforms baselines on our experiments from image recognition and sentiment analysis domains.,"Uncertainty estimation in a single forward pass without additional learnable parameters.A new method for computing output uncertainty estimates in DNNs for classification problems that matches state-of-the-art methods for uncertainty estimation and outperforms them in out-of-distribution detection tasks.The authors present inhibited softmax, a modification of the softmax through adding a constant activation which provides a measure for uncertainty. "
291,Deep Learning Inferences with Hybrid Homomorphic Encryption,"When deep learning is applied to sensitive data sets, many privacy-related implementation issues arise.These issues are especially evident in the healthcare, finance, law and government industries.Homomorphic encryption could allow a server to make inferences on inputs encrypted by a client, but to our best knowledge, there has been no complete implementation of common deep learning operations, for arbitrary model depths, using homomorphic encryption.This paper demonstrates a novel approach, efficiently implementing many deep learning functions with bootstrapped homomorphic encryption.As part of our implementation, we demonstrate Single and Multi-Layer Neural Networks, for the Wisconsin Breast Cancer dataset, as well as a Convolutional Neural Network for MNIST.Our results give promising directions for privacy-preserving representation learning, and the return of data control to users.","We made a feature-rich system for deep learning with encrypted inputs, producing encrypted outputs, preserving privacy.A framework for private deep learning model inference using FHE schemes that support fast bootstrapping and thus can reduce computation time.The paper presents a means of evaluating a neural network securely using homomorphic encryption."
292,GamePad: A Learning Environment for Theorem Proving,"In this paper, we introduce a system called GamePad that can be used to explore the application of machine learning methods to theorem proving in the Coq proof assistant.Interactive theorem provers such as Coq enable users to construct machine-checkable proofs in a step-by-step manner.Hence, they provide an opportunity to explore theorem proving with human supervision.We use GamePad to synthesize proofs for a simple algebraic rewrite problem and train baseline models for a formalization of the Feit-Thompson theorem.We address position evaluation and tactic prediction tasks, which arise naturally in tactic-based theorem proving.","We introduce a system called GamePad to explore the application of machine learning methods to theorem proving in the Coq proof assistant.This paper describes a system for applying machine learning to interactive theorem proving, focuses on tasks of tactic prediction and position evaluation, and shows that a neural model outperforms an SVM on both tasks.Proposes that machine learning techniques be used to help build proof in the theorem prover Coq."
293,Analysis of Quantized Models,"Deep neural networks are usually huge, which significantly limits the deployment on low-end devices.In recent years, manyweight-quantized models have been proposed.They have small storage and fast inference, but training can still be time-consuming.This can be improved with distributed learning.To reduce the high communication cost due to worker-server synchronization, recently gradient quantization has also been proposed to train deep networks with full-precision weights.In this paper, we theoretically study how the combination of both weight and gradient quantization affects convergence.We show that weight-quantized models converge to an error related to the weight quantization resolution and weight dimension; quantizing gradients slows convergence by a factor related to the gradient quantization resolution and dimension; and clipping the gradient before quantization renders this factor dimension-free, thus allowing the use of fewer bits for gradient quantization.Empirical experiments confirm the theoretical convergence results, and demonstrate that quantized networks can speed up training and have comparable performance as full-precision networks.","In this paper, we studied efficient training of loss-aware weight-quantized networks with quantized gradient in a distributed environment, both theoretically and empirically.This paper studies convergence properties of loss-aware weight quantization with different gradient precisions in the distributed environment, and provides convergence analysis for weight quantization with full-precision, quantized and quantized clipped gradients.The authors proposes an analysis of the effect of simultaneously quantizing the weights and gradients in training a parametrized model in a fully-synchronized distributed environment."
294,Selfless Sequential Learning,"Sequential learning, also called lifelong learning, studies the problem of learning tasks in a sequence with access restricted to only the data of the current task.In this paper we look at a scenario with fixed model capacity, and postulate that the learning process should not be selfish, i.e. it should account for future tasks to be added and thus leave enough capacity for them.To achieve Selfless Sequential Learning we study different regularization strategies and activation functions.We find thatimposing sparsity at the level of the representation is more beneficial for sequential learning than encouraging parameter sparsity.In particular, we propose a novel regularizer, that encourages representation sparsity by means of neural inhibition.It results in few active neurons which in turn leaves more free neurons to be utilized by upcoming tasks.As neural inhibition over an entire layer can be too drastic, especially for complex tasks requiring strong representations,our regularizer only inhibits other neurons in a local neighbourhood, inspired by lateral inhibition processes in the brain.We combine our novel regularizer with state-of-the-art lifelong learning methods that penalize changes to important previously learned parts of the network.We show that our new regularizer leads to increased sparsity which translates in consistent performance improvement on diverse datasets.","A regularization strategy for improving the performance of sequential learningA novel, regularization based approach to the sequential learning problem using a fixed size model that adds extra terms to the loss, encouraging representation sparsity and combating catastrophic forgetting.This paper deals with the problem of catastrophic forgetting in lifelong learning by proposing regularized learning strategies"
295,A Synaptic Neural Network and Synapse Learning,"A Synaptic Neural Network consists of synapses and neurons.Inspired by the synapse research of neuroscience, we built a synapse model with a nonlinear synapse function of excitatory and inhibitory channel probabilities.Introduced the concept of surprisal space and constructed a commutative diagram, we proved that the inhibitory probability function -log) in surprisal space is the topologically conjugate function of the inhibitory complementary probability 1-x in probability space.Furthermore, we found that the derivative of the synapse over the parameter in the surprisal space is equal to the negative Bose-Einstein distribution.In addition, we constructed a fully connected synapse graph as a synapse block of a synaptic neural network.Moreover, we proved the gradient formula of a cross-entropy loss function over parameters, so synapse learning can work with the gradient descent and backpropagation algorithms.In the proof-of-concept experiment, we performed an MNIST training and testing on the MLP model with synapse network as hidden layers.",A synaptic neural network with synapse graph and learning that has the feature of topological conjugation and Bose-Einstein distribution in surprisal space. The authors propose a hybrid neural nework composed of a synapse graph that can be embedded into a standard neural networkPresents a biologically-inspired neural network model based on the excitatory and inhibitory ion channels in the membranes of real cells
296,Generalized Graph Embedding Models,"Many types of relations in physical, biological, social and information systems can be modeled as homogeneous or heterogeneous concept graphs.Hence, learning from and with graph embeddings has drawn a great deal of research interest recently, but only ad hoc solutions have been obtained this far.In this paper, we conjecture that the one-shot supervised learning mechanism is a bottleneck in improving the performance of the graph embedding learning algorithms, and propose to extend this by introducing a multi-shot unsupervised learning framework.Empirical results on several real-world data set show that the proposed model consistently and significantly outperforms existing state-of-the-art approaches on knowledge base completion and graph based multi-label classification tasks.","Generalized Graph Embedding ModelsA generalized knowledge graph embedding approach which learns the embeddings based on three different simultaneous objectives, and performs on par or even outperforms existing state-of-the art approaches.Tackles the task of learning embeddings of multi-relational graphs using a neural networkProposes a new method, GEN, to compute embeddings of multirelationship graphs, particularly that so-called E-Cells and R-Cells can answer queries of the form (h,r,?),(?r,t), and (h,?,t)"
297,Minimax Curriculum Learning: Machine Teaching with Desirable Difficulties and Scheduled Diversity,"We introduce and study minimax curriculum learning, a new method for adaptively selecting a sequence of training subsets for a succession of stages in machine learning.The subsets are encouraged to be small and diverse early on, and then larger, harder, and allowably more homogeneous in later stages.At each stage, model weights and training sets are chosen by solving a joint continuous-discrete minimax optimization, whose objective is composed of a continuous loss and a discrete submodular promoter of diversity for the chosen subset.MCL repeatedly solves a sequence of such optimizations with a schedule of increasing training set size and decreasing pressure on diversity encouragement.We reduce MCL to the minimization of a surrogate function handled by submodular maximization and continuous gradient methods.We show that MCL achieves better performance and, with a clustering trick, uses fewer labeled samples for both shallow and deep models while achieving the same performance.Our method involves repeatedly solving constrained submodular maximization of an only slowly varying function on the same ground set.Therefore, we develop a heuristic method that utilizes the previous submodular maximization solution as a warm start for the current submodular maximization process to reduce computation while still yielding a guarantee.",Minimax Curriculum Learning is a machine teaching method involving increasing desirable hardness and scheduled reducing diversity. A curriculum learning approach using a submodular set function that captures the diversity of examples chosen during training. The paper introduces MiniMax Curriculum learning as an approach for adaptively training models by providing it different subsets of data. 
298,Implicit Causal Models for Genome-wide Association Studies,"Progress in probabilistic generative models has accelerated, developing richer models with neural architectures, implicit densities, and with scalable algorithms for their Bayesian inference.However, there has been limited progress in models that capture causal relationships, for example, how individual genetic factors cause major human diseases.In this work, we focus on two challenges in particular: How do we build richer causal models, which can capture highly nonlinear relationships and interactions between multiple causes?How do we adjust for latent confounders, which are variables influencing both cause and effect and which prevent learning of causal relationships?To address these challenges, we synthesize ideas from causality and modern probabilistic modeling.For the first, we describe implicit causal models, a class of causal models that leverages neural architectures with an implicit density.For the second, we describe an implicit causal model that adjusts for confounders by sharing strength across examples.In experiments, we scale Bayesian inference on up to a billion genetic measurements.We achieve state of the art accuracy for identifying causal factors: we significantly outperform the second best result by an absolute difference of 15-45.3%.","Implicit models applied to causality and geneticsThe authors propose to use the implicit model to tackle Genome-Wide Association problem.This paper proposes solutions for the problems in genome-wide association studies of confounding due to population structure and the potential presence of non-linear interactions between different parts of the genome, and bridges statistical genetics and ML.Presents a non-linear generative model for GWAS that models population structure where non-linearities are modeled using neural networks as non-linear function approximators and inference is performed using likelihood-free variational inference"
299,Few-Shot Learning by Exploiting Object Relation,"Few-shot learning trains image classifiers over datasets with few examples per category.It poses challenges for the optimization algorithms, which typically require many examples to fine-tune the model parameters for new categories.Distance-learning-based approaches avoid the optimization issue by embedding the images into a metric space and applying the nearest neighbor classifier for new categories.In this paper, we propose to exploit the object-level relation to learn the image relation feature, which is converted into a distance directly.For a new category, even though its images are not seen by the model, some objects may appear in the training images.Hence, object-level relation is useful for inferring the relation of images from unseen categories.Consequently, our model generalizes well for new categories without fine-tuning.Experimental results on benchmark datasets show that our approach outperforms state-of-the-art methods.",Few-shot learning by exploiting the object-level relation to learn the image-level relation (similarity)This paper deals with the problem of few-shot learning by proposing an embedding-based approach that learns to compare object-level features between support and query set examplesProposes a few shot learning method that exploits the object-level relation between different images based on neared neighbor search and concatenates feature maps of two input images into one feature map
300,The Effectiveness of Pre-Trained Code Embeddings,"Word embeddings are widely used in machine learning based natural language processing systems.It is common to use pre-trained word embeddings which provide benefits such as reduced training time and improved overall performance.There has been a recent interest in applying natural language processing techniques to programming languages.However, none of this recent work uses pre-trained embeddings on code tokens.Using extreme summarization as the downstream task, we show that using pre-trained embeddings on code tokens provides the same benefits as it does to natural languages, achieving: over 1.9x speedup, 5% improvement in test loss, 4% improvement in F1 scores, and resistance to over-fitting.We also show that the choice of language used for the embeddings does not have to match that of the task to achieve these benefits and that even embeddings pre-trained on human languages provide these benefits to programming languages. ","Researchers exploring natural language processing techniques applied to source code are not using any form of pre-trained embeddings, we show that they should be.This paper sets to understand whether pretraining word embeddings for programming language code by using NLP-like language models has an impact on extreme code summarization task.This work shows how pre-training word vectors using corpuses of code leads to representations that are more suitable than randomly initialized and trained representations for function/method name prediction"
301,Solving the Rubik's Cube with Approximate Policy Iteration,"Recently, Approximate Policy Iteration algorithms have achieved super-human proficiency in two-player zero-sum games such as Go, Chess, and Shogi without human data.These API algorithms iterate between two policies: a slow policy, and a fast policy.In these two-player games, a reward is always received at the end of the game.However, the Rubik’s Cube has only a single solved state, and episodes are not guaranteed to terminate.This poses a major problem for these API algorithms since they rely on the reward received at the end of the game.We introduce Autodidactic Iteration: an API algorithm that overcomes the problem of sparse rewards by training on a distribution of states that allows the reward to propagate from the goal state to states farther away.Autodidactic Iteration is able to learn how to solve the Rubik’s Cube and the 15-puzzle without relying on human data.Our algorithm is able to solve 100% of randomly scrambled cubes while achieving a median solve length of 30 moves — less than or equal to solvers that employ human domain knowledge.","We solve the Rubik's Cube with pure reinforcement learning"", 'Solution to solving Rubik cube using reinforcement learning (RL) with Monte-Carlo tree search (MCTS) through autodidactic iteration. This work solves Rubik's Cube using an approximate policy iteration method called Autodidactic iteration, overcoming the problem of sparse rewards by creating its own rewards system."", ""Introduces a deep RL algorithm to solve the Rubik's cube that handles the huge state space and very sparse reward of the Rubik's cube"
302,Neural Compositional Denotational Semantics for Question Answering,"Answering compositional questions requiring multi-step reasoning is challenging for current models.We introduce an end-to-end differentiable model for interpreting questions, which is inspired by formal approaches to semantics.Each span of text is represented by a denotation in a knowledge graph, together with a vector that captures ungrounded aspects of meaning.Learned composition modules recursively combine constituents, culminating in a grounding for the complete sentence which is an answer to the question.For example, to interpret ‘not green’, the model will represent ‘green’ as a set of entities, ‘not’ as a trainable ungrounded vector, and then use this vector to parametrize a composition function to perform a complement operation.For each sentence, we build a parse chart subsuming all possible parses, allowing the model to jointly learn both the composition operators and output structure by gradient descent.We show the model can learn to represent a variety of challenging semantic operators, such as quantifiers, negation, disjunctions and composed relations on a synthetic question answering task.The model also generalizes well to longer sentences than seen in its training data, in contrast to LSTM and RelNet baselines.We will release our code.","We describe an end-to-end differentiable model for QA that learns to represent spans of text in the question as denotations in knowledge graph, by learning both neural modules for composition and the syntactic structure of the sentence.This paper presents a model for visual question answering that can learn both parameters and structure predictors for a modular neural network, without supervised structures or assistance from a syntactic parser.Proposes for training a question answering model from answers only and a KB by learning latent trees that capture the syntax and learn the semantic of words"
303,DLVM: A modern compiler infrastructure for deep learning systems,"Deep learning software demands reliability and performance.However, many of the existing deep learning frameworks are software libraries that act as an unsafe DSL in Python and a computation graph interpreter.We present DLVM, a design and implementation of a compiler infrastructure with a linear algebra intermediate representation, algorithmic differentiation by adjoint code generation, domain- specific optimizations and a code generator targeting GPU via LLVM.Designed as a modern compiler infrastructure inspired by LLVM, DLVM is more modular and more generic than existing deep learning compiler frameworks, and supports tensor DSLs with high expressivity.With our prototypical staged DSL embedded in Swift, we argue that the DLVM system enables a form of modular, safe and performant frameworks for deep learning.","We introduce a novel compiler infrastructure that addresses shortcomings of existing deep learning frameworks.Proposal to move from ad-hoc code generation in deep learning engines to compiler and languages best practices.This paper presents a compiler framework that allows definition of domain-specific languages for deep learning systems, and defines compilation stages that can take advantage of standard optimizations and specialized optimizations for neural networks.This paper introduces a DLVM to take advantage of the compiler aspects of a tensor compiler"
304,Learning to navigate by distilling visual information and natural language instructions,"In this work, we focus on the problem of grounding language by training an agentto follow a set of natural language instructions and navigate to a target objectin a 2D grid environment.The agent receives visual information through rawpixels and a natural language instruction telling what task needs to be achieved.Other than these two sources of information, our model does not have any priorinformation of both the visual and textual modalities and is end-to-end trainable.We develop an attention mechanism for multi-modal fusion of visual and textualmodalities that allows the agent to learn to complete the navigation tasks and alsoachieve language grounding.Our experimental results show that our attentionmechanism outperforms the existing multi-modal fusion mechanisms proposed inorder to solve the above mentioned navigation task.We demonstrate through thevisualization of attention weights that our model learns to correlate attributes ofthe object referred in the instruction with visual representations and also showthat the learnt textual representations are semantically meaningful as they followvector arithmetic and are also consistent enough to induce translation between instructionsin different natural languages.We also show that our model generalizeseffectively to unseen scenarios and exhibit zero-shot generalization capabilities.In order to simulate the above described challenges, we introduce a new 2D environmentfor an agent to jointly learn visual and textual modalities","Attention based architecture for language grounding via reinforcement learning in a new customizable 2D grid environment The paper tackles the problem of navigation given an instruction and proposes an approach to combine textual and visual information via an attention mechanismThis paper considers the problem of following natural language instructions given a first-person view of an a priori unknown environment, and proposes a neural architecture method.Studies the problem of navigating to a target object in a 2D grid environment by following given natural language description and receiving visual information as raw pixels."
305,QANet: Combining Local Convolution with Global Self-Attention for Reading Comprehension," Current end-to-end machine reading and question answering models are primarily based on recurrent neural networks with attention.Despite their success, these models are often slow for both training and inference due to the sequential nature of RNNs.We propose a new Q&A architecture called QANet, which does not require recurrent networks: Its encoder consists exclusively of convolution and self-attention, where convolution models local interactions and self-attention models global interactions.On the SQuAD dataset, our model is 3x to 13x faster in training and 4x to 9x faster in inference, while achieving equivalent accuracy to recurrent models.The speed-up gain allows us to train the model with much more data.We hence combine our model with data generated by backtranslation from a neural machine translation model.On the SQuAD dataset, our single model, trained with augmented data, achieves 84.6 F1 score on the test set, which is significantly better than the best published F1 score of 81.8.",A simple architecture consisting of convolutions and attention achieves results on par with the best documented recurrent models.A fast high performance paraphrasing based data augmentation method and a non-recurrent reading comprehension model using only convolutions and attention.This paper proposes applying CNNs+self-attention modules instead of LSTMs and enhancing the RC model training with passage paraphrases generated by a neural paraphrasing model in order to improve RC performance.This paper presents a reading comprehension model using convolutions and attention and propose to augment additional training data by paraphrasing based on off-the-shelf neural machine translation
306,Spherical CNNs,"Convolutional Neural Networks have become the method of choice for learning problems involving 2D planar images.However, a number of problems of recent interest have created a demand for models that can analyze spherical images.Examples include omnidirectional vision for drones, robots, and autonomous cars, molecular regression problems, and global weather and climate modelling.A naive application of convolutional networks to a planar projection of the spherical signal is destined to fail, because the space-varying distortions introduced by such a projection will make translational weight sharing ineffective.In this paper we introduce the building blocks for constructing spherical CNNs.We propose a definition for the spherical cross-correlation that is both expressive and rotation-equivariant.The spherical correlation satisfies a generalized Fourier theorem, which allows us to compute it efficiently using a generalized Fast Fourier Transform algorithm.We demonstrate the computational efficiency, numerical accuracy, and effectiveness of spherical CNNs applied to 3D model recognition and atomization energy regression.","We introduce Spherical CNNs, a convolutional network for spherical signals, and apply it to 3D model recognition and molecular energy regression.The paper proposes a framework for constructing spherical convolutional networks based on a novel synthesis of several existing conceptsThis paper focuses on how to extend convolutional neural networks to have built-in spherical invariance, and adapts tools from non-Abelian harmonic analysis to achieve this goal.The authors develop a novel scheme for representing spherical data from the ground up"
307,AUTOMATED DESIGN USING NEURAL NETWORKS AND GRADIENT DESCENT,"We propose a novel method that makes use of deep neural networks and gradient decent to perform automated design on complex real world engineering tasks.Our approach works by training a neural network to mimic the fitness function of a design optimization task and then, using the differential nature of the neural network, perform gradient decent to maximize the fitness.We demonstrate this methods effectiveness by designing an optimized heat sink and both 2D and 3D airfoils that maximize the lift drag ratio under steady state flow conditions.We highlight that our method has two distinct benefits over other automated design approaches.First, evaluating the neural networks prediction of fitness can be orders of magnitude faster then simulating the system of interest.Second, using gradient decent allows the design space to be searched much more efficiently then other gradient free methods.These two strengths work together to overcome some of the current shortcomings of automated design.","A method for performing automated design on real world objects such as heat sinks and wing airfoils that makes use of neural networks and gradient descent.Neural network (parameterization and prediction) and gradient descent (back propogation) to automatically design for engineering tasks. This paper introduces using a deep network to approximate the behavior of a complex physical system, and then design optimal devices by optimizing this network with respect to its inputs."
308,Stable Distribution Alignment Using the Dual of the Adversarial Distance,"Methods that align distributions by minimizing an adversarial distance between them have recently achieved impressive results.However, these approaches are difficult to optimize with gradient descent and they often do not converge well without careful hyperparameter tuning and proper initialization.We investigate whether turning the adversarial min-max problem into an optimization problem by replacing the maximization part with its dual improves the quality of the resulting alignment and explore its connections to Maximum Mean Discrepancy.Our empirical results suggest that using the dual formulation for the restricted family of linear discriminators results in a more stable convergence to a desirable solution when compared with the performance of a primal min-max GAN-like objective and an MMD objective under the same restrictions.We test our hypothesis on the problem of aligning two synthetic point clouds on a plane and on a real-image domain adaptation problem on digits.In both cases, the dual formulation yields an iterative procedure that gives more stable and monotonic improvement over time."," We propose a dual version of the logistic adversarial distance for feature alignment and show that it yields more stable gradient step iterations than the min-max objective.The paper deals with fixing GANs at the computational levelThis paper studies a dual formulation of an adversarial loss based on an upper-bound of the logistic loss, and turns the standard min max problem of adversarial training into a single minimization problem.Proposes to re-formulate the GAN saddle point objective (for a logistic regression discriminator) as a minimization problem by dualizing the maximum likelihood objective for regularized logistic regression"
309,Espresso: Efficient Forward Propagation for Binary Deep Neural Networks," There are many applications scenarios for which the computational performance and memory footprint of the prediction phase of Deep Neural Networks need to be optimized.Binary Deep Neural Networks have been shown to be an effective way of achieving this objective.In this paper, we show how Convolutional Neural Networks can be implemented using binary representations.Espresso is a compact, yet powerful library written in C/CUDA that features all the functionalities required for the forward propagation of CNNs, in a binary file less than 400KB, without any external dependencies.Although it is mainly designed to take advantage of massive GPU parallelism, Espresso also provides an equivalent CPU implementation for CNNs.Espresso provides special convolutional and dense layers for BCNNs, leveraging bit-packing and bit-wise computations for efficient execution.These techniques provide a speed-up of matrix-multiplication routines, and at the same time, reduce memory usage when storing parameters and activations.We experimentally show that Espresso is significantly faster than existing implementations of optimized binary neural networks.Espresso is released under the Apache 2.0 license and is available at http://github.com/organization/project.","state-of-the-art computational performance implementation of binary neural networksThe paper presents a library written in C/CUDA that features all the functionalities required for the forward propagation of BCNNsThis paper builds on Binary-NET and expands it to CNN architectures, provides optimizations that improve the speed of the forward pass, and provides optimized code for Binary CNN."
310,Differentiable Greedy Networks,"Optimal selection of a subset of items from a given set is a hard problem that requires combinatorial optimization.In this paper, we propose a subset selection algorithm that is trainable with gradient based methods yet achieves near optimal performance via submodular optimization.We focus on the task of identifying a relevant set of sentences for claim verification in the context of the FEVER task.Conventional methods for this task look at sentences on their individual merit and thus do not optimize the informativeness of sentences as a set.We show that our proposed method which builds on the idea of unfolding a greedy algorithm into a computational graph allows both interpretability and gradient based training.The proposed differentiable greedy network outperforms discrete optimization algorithms as well as other baseline methods in terms of precision and recall.","We propose a subset selection algorithm that is trainable with gradient based methods yet achieves near optimal performance via submodular optimization. Proposes a neural network based model that integrates submodular function by combining gradient based optimization technique with submodular framework named 'Differentiable Greedy Network' (DGN)."", 'Proposes a neural network that aims to select a subset of elements (e.g. selecting k sentences that are mostly related to a claim from a set of retrieved docs)"
311,Hierarchically Clustered Representation Learning,"The joint optimization of representation learning and clustering in the embedding space has experienced a breakthrough in recent years.In spite of the advance, clustering with representation learning has been limited to flat-level categories, which oftentimes involves cohesive clustering with a focus on instance relations.To overcome the limitations of flat clustering, we introduce hierarchically clustered representation learning, which simultaneously optimizes representation learning and hierarchical clustering in the embedding space.Specifically, we place a nonparametric Bayesian prior on embeddings to handle dynamic mixture hierarchies under the variational autoencoder framework, and to adopt the generative process of a hierarchical-versioned Gaussian mixture model.Compared with a few prior works focusing on unifying representation learning and hierarchical clustering, HCRL is the first model to consider a generation of deep embeddings from every component of the hierarchy, not just leaf components.This generation process enables more meaningful separations and mergers of clusters via branches in a hierarchy.In addition to obtaining hierarchically clustered embeddings, we can reconstruct data by the various abstraction levels, infer the intrinsic hierarchical structure, and learn the level-proportion features.We conducted evaluations with image and text domains, and our quantitative analyses showed competent likelihoods and the best accuracies compared with the baselines.","We introduce hierarchically clustered representation learning (HCRL), which simultaneously optimizes representation learning and hierarchical clustering in the embedding space.The paper proposes using the nested CRP as a clustering model rather than a topic modelPresents a novel hierarchical clustering method over an embedding space where both the embedding space and the heirarchical clustering are simultaneously learnt"
312,GEOMETRIC AUGMENTATION FOR ROBUST NEURAL NETWORK CLASSIFIERS,"We introduce a novel geometric perspective and unsupervised model augmentation framework for transforming traditional deep neural networks into adversarially robust classifiers.Class-conditional probability densities based on Bayesian nonparametric mixtures of factor analyzers over the input space are used to design soft decision labels for feature to label isometry.Classconditional distributions over features are also learned using BNP-MFA to develop plug-in maximum a posterior classifiers to replace the traditional multinomial logistic softmax classification layers.This novel unsupervised augmented framework, which we call geometrically robust networks, is applied to CIFAR-10, CIFAR-100, and to Radio-ML.We demonstrate the robustness of GRN models to adversarial attacks from fast gradient sign method, Carlini-Wagner, and projected gradient descent.",We develop a statistical-geometric unsupervised learning augmentation framework for deep neural networks to make them robust to adversarial attacks.Transfroms traditional deep neural networks into adversarial robust calssifiers using GRNsProposes a defense based on class-conditional feature distributions to turn deep neural netwroks into robust classifiers
313,Structured Exploration via Hierarchical Variational Policy Networks,"Reinforcement learning in environments with large state-action spaces is challenging, as exploration can be highly inefficient.Even if the dynamics are simple, the optimal policy can be combinatorially hard to discover.In this work, we propose a hierarchical approach to structured exploration to improve the sample efficiency of on-policy exploration in large state-action spaces.The key idea is to model a stochastic policy as a hierarchical latent variable model, which can learn low-dimensional structure in the state-action space, and to define exploration by sampling from the low-dimensional latent space.This approach enables lower sample complexity, while preserving policy expressivity.In order to make learning tractable, we derive a joint learning and exploration strategy by combining hierarchical variational inference with actor-critic learning.The benefits of our learning approach are that1) it is principled,2) simple to implement,3) easily scalable to settings with many actions and4) easily composable with existing deep learning approaches.We demonstrate the effectiveness of our approach on learning a deep centralized multi-agent policy, as multi-agent environments naturally have an exponentially large state-action space.In this setting, the latent hierarchy implements a form of multi-agent coordination during exploration and execution.We demonstrate empirically that MACE can more efficiently learn optimal policies in challenging multi-agent games with a large number of agents, compared to conventional baselines.Moreover, we show that our hierarchical structure leads to meaningful agent coordination.","Make deep reinforcement learning in large state-action spaces more efficient using structured exploration with deep hierarchical policies. A method to coordinate agent behaviour by using policies that have shared latent structure, a variational policy optimization method to optimize the coordinated policies, and a derivation of the authors' variational, hierarchical update."", 'This paper suggests an algorithmic innovation consisting of hierarchical latent variables for coordinated exploration in multi-agent settings"
314,Revealing interpretable object representations from human behavior,"To study how mental object representations are related to behavior, we estimated sparse, non-negative representations of objects using human behavioral judgments on images representative of 1,854 object categories.These representations predicted a latent similarity structure between objects, which captured most of the explainable variance in human behavioral judgments.Individual dimensions in the low-dimensional embedding were found to be highly reproducible and interpretable as conveying degrees of taxonomic membership, functionality, and perceptual attributes.We further demonstrated the predictive power of the embeddings for explaining other forms of human behavior, including categorization, typicality judgments, and feature ratings, suggesting that the dimensions reflect human conceptual representations of objects beyond the specific task.","Human behavioral judgments are used to obtain sparse and interpretable representations of objects that generalize to other tasksThis paper describes a large-scale experiment on human object/sematic representations and a model of such representations.This paper develops a new representation system for object representations from training on data collected from odd-one-out human judgements of images.A new approach to learn a sparse, positive, interpretable semantic space that maximizes human similarity judgements by training to specifically maximize the prediction of human similarity judgements."
315,Ask the Right Questions: Active Question Reformulation with Reinforcement Learning,"We frame Question Answering as a Reinforcement Learning task, an approach that we call Active Question Answering.We propose an agent that sits between the user and a black box QA system and learns to reformulate questions to elicit the best possible answers.The agent probes the system with, potentially many, natural language reformulations of an initial question and aggregates the returned evidence to yield the best answer.The reformulation system is trained end-to-end to maximize answer quality using policy gradient.We evaluate on SearchQA, a dataset of complex questions extracted from Jeopardy!.The agent outperforms a state-of-the-art base model, playing the role of the environment, and other benchmarks.We also analyze the language that the agent has learned while interacting with the question answering system. We find that successful question reformulations look quite different from natural language paraphrases. The agent is able to discover non-trivial reformulation strategies that resemble classic information retrieval techniques such as term re-weighting and stemming.",We propose an agent that sits between the user and a black box question-answering system and which learns to reformulate questions to elicit the best possible answersThis paper proposes active question answering via a reinforcement learning approach that learns to rephrase questions in a way to provide the best possible answers.Clearly describes how the researchers designed and actively trained two models for question reformulation and answer selection during question answering episodes
316,Learning Priors for Adversarial Autoencoders,"Most deep latent factor models choose simple priors for simplicity, tractabilityor not knowing what prior to use.Recent studies show that the choice ofthe prior may have a profound effect on the expressiveness of the model,especially when its generative network has limited capacity.In this paper, we propose to learn a proper prior from data for adversarial autoencoders.We introduce the notion of code generators to transform manually selectedsimple priors into ones that can better characterize the data distribution.Experimental results show that the proposed model can generate better image quality and learn better disentangled representations thanAAEs in both supervised and unsupervised settings.Lastly, we present itsability to do cross-domain translation in a text-to-image synthesis task.",Learning Priors for Adversarial AutoencodersProposes a simple extension of adversarial auto-encoders for conditional image generation.Focuses on adversarial autoencoders and introduces a code generator network to transform a simple prior into one that together with the generator can better fit the data distribution
317, Generating Text through Adversarial Training using Skip-Thought Vectors,"In the past few years, various advancements have been made in generative models owing to the formulation of Generative Adversarial Networks.GANs have been shown to perform exceedingly well on a wide variety of tasks pertaining to image generation and style transfer.In the field of Natural Language Processing, word embeddings such as word2vec and GLoVe are state-of-the-art methods for applying neural network models on textual data.Attempts have been made for utilizing GANs with word embeddings for text generation.This work presents an approach to text generation using Skip-Thought sentence embeddings in conjunction with GANs based on gradient penalty functions and f-measures.The results of using sentence embeddings with GANs for generating text conditioned on input information are comparable to the approaches where word embeddings are used.",Generating text using sentence embeddings from Skip-Thought Vectors with the help of Generative Adversarial Networks.Describes application of generative adversarial networks for modeling textual data with the help of ski-thought vectors and experiments with different flavors of GANs for two different datasets.
318,Unbiased Online Recurrent Optimization,"The novel algorithm allows for online learning of general recurrent computational graphs such as recurrent network models.It works in a streaming fashion and avoids backtracking through past activations and inputs.UORO is computationally as costly as , a widespread algorithm for online learning of recurrent networks . UORO is a modification of that bypasses the need for model sparsity and makes implementation easy in current deep learning frameworks, even for complex models. Like NoBackTrack, UORO provides unbiased gradient estimates; unbiasedness is the core hypothesis in stochastic gradient descent theory, without which convergence to a local optimum is not guaranteed.On the contrary, truncated BPTT does not provide this property, leading to possible divergence. On synthetic tasks where truncated BPTT is shown to diverge, UORO converges.For instance, when a parameter has a positive short-term but negative long-term influence, truncated BPTT diverges unless the truncation span is very significantly longer than the intrinsic temporal range of the interactions, while UORO performs well thanks to the unbiasedness of its gradients.","Introduces an online, unbiased and easily implementable gradient estimate for recurrent models.The authors introduce a novel approach to online learning of the parameters of recurrent neural networks from long sequences that overcomes the imitation of truncated backpropagation through timeThis paper approaches online training of RNNs in a principled way, and proposes a modification to RTRL and to use forward approach for gradient calculation."
319,Label super-resolution networks,"We present a deep learning-based method for super-resolving coarse labels assigned to groups of image pixels into pixel-level labels, given the joint distribution between those low- and high-resolution labels.This method involves a novel loss function that minimizes the distance between a distribution determined by a set of model outputs and the corresponding distribution given by low-resolution labels over the same set of outputs.This setup does not require that the high-resolution classes match the low-resolution classes and can be used in high-resolution semantic segmentation tasks where high-resolution labeled data is not available.Furthermore, our proposed method is able to utilize both data with low-resolution labels and any available high-resolution labels, which we show improves performance compared to a network trained only with the same amount of high-resolution data.We test our proposed algorithm in a challenging land cover mapping task to super-resolve labels at a 30m resolution to a separate set of labels at a 1m resolution.We compare our algorithm with models that are trained on high-resolution data and show that1) we can achieve similar performance using only low-resolution data; and2) we can achieve better performance when we incorporate a small amount of high-resolution data in our training.We also test our approach on a medical imaging problem, resolving low-resolution probability maps into high-resolution segmentation of lymphocytes with accuracy equal to that of fully supervised models.","Super-resolving coarse labels into pixel-level labels, applied to aerial imagery and medical scans.A method to super-resolve coarse low-res segmentation labels if the joint distribution of low-res and high-res labels are known."
320,Manifold Alignment via Feature Correspondence,"We propose a novel framework for combining datasets via alignment of their associated intrinsic dimensions.Our approach assumes that the two datasets are sampled from a common latent space, i.e., they measure equivalent systems.Thus, we expect there to exist a natural alignment of the data manifolds associated with the intrinsic geometry of these datasets, which are perturbed by measurement artifacts in the sampling process.Importantly, we do not assume any individual correspondence between data points.Instead, we rely on our assumption that a subset of data features have correspondence across datasets.We leverage this assumption to estimate relations between intrinsic manifold dimensions, which are given by diffusion map coordinates over each of the datasets.We compute a correlation matrix between diffusion coordinates of the datasets by considering graph Fourier coefficients of corresponding data features.We then orthogonalize this correlation matrix to form an isometric transformation between the diffusion maps of the datasets.Finally, we apply this transformation to the diffusion coordinates and construct a unified diffusion geometry of the datasets together.We show that this approach successfully corrects misalignment artifacts, and allows for integrated data.",We propose a method for aligning the latent features learned from different datasets using harmonic correlations.Proposes using feature correspondences to preform manifold alignment between batches of data from the same samples to avoid the collection of noisy measurements.
321,The Body is not a Given: Joint Agent Policy Learning and Morphology Evolution,"Reinforcement learning has proven to be a powerful paradigm for deriving complex behaviors from simple reward signals in a wide range of environments.When applying RL to continuous control agents in simulated physics environments, the body is usually considered to be part of the environment.However, during evolution the physical body of biological organisms and their controlling brains are co-evolved, thus exploring a much larger space of actuator/controller configurations.Put differently, the intelligence does not reside only in the agent's mind, but also in the design of their body.We propose a method for uncovering strong agents, consisting of a good combination of a body and policy, based on combining RL with an evolutionary procedure.Given the resulting agent, we also propose an approach for identifying the body changes that contributed the most to the agent performance.We use the Shapley value from cooperative game theory to find the fair contribution of individual components, taking into account synergies between components.We evaluate our methods in an environment similar to the the recently proposed Robo-Sumo task, where agents in a 3D environment with simulated physics compete in tipping over their opponent or pushing them out of the arena.Our results show that the proposed methods are indeed capable of generating strong agents, significantly outperforming baselines that focus on optimizing the agent policy alone.A video is available at: www.youtube.com/watch?v=eei6Rgom3YY",Evolving the shape of the body in RL controlled agents improves their performance (and help learning)PEOM algorithm that incorporates Shapley value to accelerate the evolution by identifying contribution of each body part
322,Avoiding Catastrophic States with Intrinsic Fear,"Many practical reinforcement learning problems contain catastrophic states that the optimal policy visits infrequently or never.Even on toy problems, deep reinforcement learners periodically revisit these states, once they are forgotten under a new policy.In this paper, we introduce intrinsic fear, a learned reward shaping that accelerates deep reinforcement learning and guards oscillating policies against periodic catastrophes.Our approach incorporates a second model trained via supervised learning to predict the probability of imminent catastrophe.This score acts as a penalty on the Q-learning objective.Our theoretical analysis demonstrates that the perturbed objective yields the same average return under strong assumptions and an-close average return under weaker assumptions.Our analysis also shows robustness to classification errors.Equipped with intrinsic fear, our DQNs solve the toy environments and improve on the Atari games Seaquest, Asteroids, and Freeway.","Shape reward with intrinsic motivation to avoid catastrophic states and mitigate catastrophic forgetting.An RL algorithm that combines the DQN algorithm with a fear model trained in parallel to predict catastropohic states.The paper studies catastrophic forgetting in RL, by emphasizing tasks where a DQN is able to learn to avoid catastrophic events as long as it avoids forgetting."
323,Volumetric Convolution: Automatic Representation Learning in Unit Ball,"Convolution is an efficient technique to obtain abstract feature representations using hierarchical layers in deep networks.Although performing convolution in Euclidean geometries is fairly straightforward, its extension to other topological spaces such as a sphere S^2 or a unit ball B^3 entails unique challenges.In this work, we propose a novel `""volumetric convolution"" operation that can effectively convolve arbitrary functions in B^3.We develop a theoretical framework for ""volumetric convolution"" based on Zernike polynomials and efficiently implement it as a differentiable and an easily pluggable layer for deep networks.Furthermore, our formulation leads to derivation of a novel formula to measure the symmetry of a function in B^3 around an arbitrary axis, that is useful in 3D shape analysis tasks.We demonstrate the efficacy of proposed volumetric convolution operation on a possible use-case i.e., 3D object recognition task.",A novel convolution operator for automatic representation learning inside unit ballThis work is related to the recent spherical CNN and SE(n) equivariant network papers and extends previous ideas to volumetric data in the unit ball.Proposes using volumetric convolutions on convolutions networks in order to learn unit ball and discusses methodology and results of process.
324,Learning to Navigate the Web,"Learning in environments with large state and action spaces, and sparse rewards, can hinder a Reinforcement Learning agent’s learning through trial-and-error.For instance, following natural language instructions on the Web leads to RL settings where input vocabulary and number of actionable elements on a page can grow very large.Even though recent approaches improve the success rate on relatively simple environments with the help of human demonstrations to guide the exploration, they still fail in environments where the set of possible instructions can reach millions.We approach the aforementioned problems from a different perspective and propose guided RL approaches that can generate unbounded amount of experience for an agent to learn from.Instead of learning from a complicated instruction with a large vocabulary, we decompose it into multiple sub-instructions and schedule a curriculum in which an agent is tasked with a gradually increasing subset of these relatively easier sub-instructions.In addition, when the expert demonstrations are not available, we propose a novel meta-learning framework that generates new instruction following tasks and trains the agent more effectively.We train DQN, deep reinforcement learning agent, with Q-value function approximated with a novel QWeb neural network architecture on these smaller, synthetic instructions.We evaluate the ability of our agent to generalize to new instructions onWorld of Bits benchmark, on forms with up to 100 elements, supporting 14 million possible instructions.The QWeb agent outperforms the baseline without using any human demonstration achieving 100% success rate on several difficult environments.","We train reinforcement learning policies using reward augmentation, curriculum learning, and meta-learning to successfully navigate web pages.Develops a curriculum learning method for training an RL agent to navigate a web, based on the idea of decomposing an instruction in to multiple sub-instructions."
325,Zero-shot Cross Language Text Classification,"Labeled text classification datasets are typically only available in a few select languages.In order to train a model for e.g news categorization in a language without a suitable text classification dataset there are two options.The first option is to create a new labeled dataset by hand, and the second option is to transfer label information from an existing labeled dataset in a source language to the target language. In this paper we propose a method for sharing label information across languages by means of a language independent text encoder.The encoder will give almost identical representations to multilingual versions of the same text.This means that labeled data in one language can be used to train a classifier that works for the rest of the languages.The encoder is trained independently of any concrete classification task and can therefore subsequently be used for any classification task. We show that it is possible to obtain good performance even in the case where only a comparable corpus of texts is available.",Cross Language Text Classification by universal encodingThis paper proposes an approach to cross-lingual text classification through the use of comparable corpora.Learning cross-lingual embeddings and training a classifier using labelled data in the source language to address learning a cross-language text categorizer with no labelled information in the target language
326,Unsupervised Latent Tree Induction with Deep Inside-Outside Recursive Auto-Encoders ,"Syntax is a powerful abstraction for language understanding.Many downstream tasks require segmenting input text into meaningful constituent chunks; more generally, models for learning semantic representations of text benefit from integrating syntax in the form of parse trees.Supervised parsers have traditionally been used to obtain these trees, but lately interest has increased in unsupervised methods that induce syntactic representations directly from unlabeled text.To this end, we propose the deep inside-outside recursive autoencoder, a fully-unsupervised method for discovering syntax that simultaneously learns representations for constituents within the induced tree.Unlike many prior approaches, DIORA does not rely on supervision from auxiliary downstream tasks and is thus not constrained to particular domains.Furthermore, competing approaches do not learn explicit phrase representations along with tree structures, which limits their applicability to phrase-based tasks.Extensive experiments on unsupervised parsing, segmentation, and phrase clustering demonstrate the efficacy of our method.DIORA achieves the state of the art in unsupervised parsing on the benchmark WSJ dataset.",In this work we propose deep inside-outside recursive auto-encoders(DIORA) a fully unsupervised method of discovering syntax while simultaneously learning representations for discovered constituents. A neural latent tree model trained with an auto-encoding objective that achieves state of the art on unsupervised constituency parsing and captures syntactic structure better than other latent tree models.The paper proposes a model for unsupervised dependency parsing (latent tree induction) that is based on a combination of the inside-outside algorithm with neural modeling (recursive auto-encoders). 
327,Understanding Short-Horizon Bias in Stochastic Meta-Optimization,"Careful tuning of the learning rate, or even schedules thereof, can be crucial to effective neural net training.There has been much recent interest in gradient-based meta-optimization, where one tunes hyperparameters, or even learns an optimizer, in order to minimize the expected loss when the training procedure is unrolled.But because the training procedure must be unrolled thousands of times, the meta-objective must be defined with an orders-of-magnitude shorter time horizon than is typical for neural net training.We show that such short-horizon meta-objectives cause a serious bias towards small step sizes, an effect we term short-horizon bias.We introduce a toy problem, a noisy quadratic cost function, on which we analyze short-horizon bias by deriving and comparing the optimal schedules for short and long time horizons.We then run meta-optimization experiments on standard benchmark datasets, showing that meta-optimization chooses too small a learning rate by multiple orders of magnitude, even when run with a moderately long time horizon typical of work in the area.We believe short-horizon bias is a fundamental problem that needs to be addressed if meta-optimization is to scale to practical neural net training regimes.",We investigate the bias in the short-horizon meta-optimization objective.This paper proposes a simplified model and problem to demonstrate the short-horizon bias of the learning rate meta-optimization.This paper studies the issue of truncated backpropagation for meta-optimization through a number of experiments on a toy problem
328,Neural Persistence: A Complexity Measure for Deep Neural Networks Using Algebraic Topology,"While many approaches to make neural networks more fathomable have been proposed, they are restricted to interrogating the network with input data.Measures for characterizing and monitoring structural properties, however, have not been developed.In this work, we propose neural persistence, a complexity measure for neural network architectures based on topological data analysis on weighted stratified graphs.To demonstrate the usefulness of our approach, we show that neural persistence reflects best practices developed in the deep learning community such as dropout and batch normalization.Moreover, we derive a neural persistence-based stopping criterion that shortens the training process while achieving comparable accuracies as early stopping based on validation loss.","We develop a new topological complexity measure for deep neural networks and demonstrate that it captures their salient properties.This paper proposes the notion of neural persistence, a topological measure to assign scores to fully-connected layers in a neural network.Paper proposes to analyze the complexity of a neural network using its zero-th persistent homology."
329,Decision Boundary Analysis of Adversarial Examples,"Deep neural networks are vulnerable to adversarial examples, which are carefully crafted instances aiming to cause prediction errors for DNNs.Recent research on adversarial examples has examined local neighborhoods in the input space of DNN models.However, previous work has limited what regions to consider, focusing either on low-dimensional subspaces or small balls.In this paper, we argue that information from larger neighborhoods, such as from more directions and from greater distances, will better characterize the relationship between adversarial examples and the DNN models.First, we introduce an attack, OPTMARGIN, which generates adversarial examples robust to small perturbations.These examples successfully evade a defense that only considers a small ball around an input instance.Second, we analyze a larger neighborhood around input instances by looking at properties of surrounding decision boundaries, namely the distances to the boundaries and the adjacent classes.We find that the boundaries around these adversarial examples do not resemble the boundaries around benign examples.Finally, we show that, under scrutiny of the surrounding decision boundaries, our OPTMARGIN examples do not convincingly mimic benign examples.Although our experiments are limited to a few specific attacks, we hope these findings will motivate new, more evasive attacks and ultimately, effective defenses.","Looking at decision boundaries around an input gives you more information than a fixed small neighborhoodThe authors present a novel attack for generating adversarial examples where they attack classifiers created by randomly classifying L2 small perturbationsA new approach to generate adversarial attacks to a neural network, and a method to defend a neural network from those attacks."
330,Applications of Gaussian Processes in Finance,"Estimating covariances between financial assets plays an important role in risk management.In practice, when the sample size is small compared to the number of variables, the empirical estimate is known to be very unstable.Here, we propose a novel covariance estimator based on the Gaussian Process Latent Variable Model.Our estimator can be considered as a non-linear extension of standard factor models with readily interpretable parameters reminiscent of market betas.Furthermore, our Bayesian treatment naturally shrinks the sample covariance matrix towards a more structured matrix given by the prior and thereby systematically reduces estimation errors.Finally, we discuss some financial applications of the GP-LVM model.","Covariance matrix estimation of financial assets with Gaussian Process Latent Variable ModelsIllustrates how the Gaussian Process Latent Variable Model (GP-LVM) can replace classical linear factor models for the estimation of covariance matrices in portfolio optimization problems.This paper uses standard GPLVMs to model the covariance structure and a latent space representation of S&P500 financial time series, to optimize portfolios and predict missing values.This paper proposes to use a GPLVM to model financial returns"
331,Variance Regularizing Adversarial Learning,"We study how, in generative adversarial networks, variance in the discriminator's output affects the generator's ability to learn the data distribution.In particular, we contrast the results from various well-known techniques for training GANs when the discriminator is near-optimal and updated multiple times per update to the generator.As an alternative, we propose an additional method to train GANs by explicitly modeling the discriminator's output as a bi-modal Gaussian distribution over the real/fake indicator variables.In order to do this, we train the Gaussian classifier to match the target bi-modal distribution implicitly through meta-adversarial training.We observe that our new method, when trained together with a strong discriminator, provides meaningful, non-vanishing gradients.","We introduce meta-adversarial learning, a new technique to regularize GANs, and propose a training method by explicitly controlling the discriminator's output distribution."", 'The paper proposes variance regularizing adversarial learning for training GANs to ensure that the gradient for the generator does not vanish"
332,Noisy Networks For Exploration,"We introduce NoisyNet, a deep reinforcement learning agent with parametric noise added to its weights, and show that the induced stochasticity of the agent’s policy can be used to aid efficient exploration.The parameters of the noise are learned with gradient descent along with the remaining network weights. NoisyNet is straightforward to implement and adds little computational overhead.We find that replacing the conventional exploration heuristics for A3C, DQN and Dueling agents with NoisyNet yields substantially higher scores for a wide range of Atari games, in some cases advancing the agent from sub to super-human performance.","A deep reinforcement learning agent with parametric noise added to its weights can be used to aid efficient exploration.This paper introduces NoisyNets, neural networks whose parameters are perturbed by a parametric noise function, that obtain substantial performance improvement over baseline deep reinforcement learning algorithms. New exploration method for deep RL by injecting noise into deep networks' weights, with the noise taking various forms"
333,Active Neural Localization,"Localization is the problem of estimating the location of an autonomous agent from an observation and a map of the environment.Traditional methods of localization, which filter the belief based on the observations, are sub-optimal in the number of steps required, as they do not decide the actions taken by the agent.We propose ""Active Neural Localizer"", a fully differentiable neural network that learns to localize efficiently.The proposed model incorporates ideas of traditional filtering-based localization methods, by using a structured belief of the state with multiplicative interactions to propagate belief, and combines it with a policy model to minimize the number of steps required for localization.Active Neural Localizer is trained end-to-end with reinforcement learning.We use a variety of simulation environments for our experiments which include random 2D mazes, random mazes in the Doom game engine and a photo-realistic environment in the Unreal game engine.The results on the 2D environments show the effectiveness of the learned policy in an idealistic setting while results on the 3D environments demonstrate the model's capability of learning the policy and perceptual model jointly from raw-pixel based RGB observations.We also show that a model trained on random textures in the Doom environment generalizes well to a photo-realistic office space environment in the Unreal engine.","Active Neural Localizer"", a fully differentiable neural network that learns to localize efficiently using deep reinforcement learning. This paper formulates the problem of localisation on a known map using a belief network as an RL problem where the agent's goal is to minimise the number of steps to localise itself."", 'This is a clear and interesting paper that builds a parameterized network to select actions for a robot in a simulated environment"
334,Hint-based Training for Non-Autoregressive Translation,"Machine translation is an important real-world application, and neural network-based AutoRegressive Translation models have achieved very promising accuracy.Due to the unparallelizable nature of the autoregressive factorization, ART models have to generate tokens one by one during decoding and thus suffer from high inference latency.Recently, Non-AutoRegressive Translation models were proposed to reduce the inference time.However, they could only achieve inferior accuracy compared with ART models.To improve the accuracy of NART models, in this paper, we propose to leverage the hints from a well-trained ART model to train the NART model.We define two hints for the machine translation task: hints from hidden states and hints from word alignments, and use such hints to regularize the optimization of NART models.Experimental results show that the NART model trained with hints could achieve significantly better translation performance than previous NART models on several tasks.In particular, for the WMT14 En-De and De-En task, we obtain BLEU scores of 25.20 and 29.52 respectively, which largely outperforms the previous non-autoregressive baselines.It is even comparable to a strong LSTM-based ART model, but one order of magnitude faster in inference.","We develop a training algorithm for non-autoregressive machine translation models, achieving comparable accuracy to strong autoregressive baselines, but one order of magnitude faster in inference. Distills knowledge from intermediary hidden states and attention weights to improve non-autoregressive neural machine translation.Proposes to leverage well trained autoregressive model to inform the hidden states and the word alignment of non-autoregressive Neural Machine Translation models."
335,Dense Morphological Network: An Universal Function Approximator,"Artificial neural networks are built on the basic operation of linear combination and non-linear activation function.Theoretically this structure can approximate any continuous function with three layer architecture.But in practice learning the parameters of such network can be hard.Also the choice of activation function can greatly impact the performance of the network.In this paper we are proposing to replace the basic linear combination operation with non-linear operations that do away with the need of additional non-linear activation function.To this end we are proposing the use of elementary morphological operations as the basic operation in neurons.We show that these networks with morphological operations can approximate any smooth function requiring less number of parameters than what is necessary for normal neural networks.The results show that our network perform favorably when compared with similar structured network.We have carried out our experiments on MNIST, Fashion-MNIST, CIFAR10 and CIFAR100.","Using mophological operation (dilation and erosion) we have defined a class of network which can approximate any continious function. This paper proposes to replace standard RELU/tanh units with a combination of dilation and erosion operations, observing that the new operator creates more hyper-planes and has more expressive power.The authors introduce Morph-Net, a single layer neural network where the mapping is performed using morphological dilation and erosion."
336,Integral Pruning on Activations and Weights for Efficient Neural Networks,"With the rapidly scaling up of deep neural networks, extensive research studies on network model compression such as weight pruning have been performed for efficient deployment.This work aims to advance the compression beyond the weights to the activations of DNNs.We propose the Integral Pruning technique which integrates the activation pruning with the weight pruning.Through the learning on the different importance of neuron responses and connections, the generated network, namely IPnet, balances the sparsity between activations and weights and therefore further improves execution efficiency.The feasibility and effectiveness of IPnet are thoroughly evaluated through various network models with different activation functions and on different datasets.With <0.5% disturbance on the testing accuracy, IPnet saves 71.1% ~ 96.35% of computation cost, compared to the original dense models with up to 5.8x and 10x reductions in activation and weight numbers, respectively.","This work advances DNN compression beyond the weights to the activations by integrating the activation pruning with the weight pruning. An integral model compression method that handles both weight and activation pruning, leading to more efficient network computation and effective reduction of the number of multiply-and-accumulate.This article presents a novel approach to reduce the computation cost of deep neural networks by integrating activation pruning along with weight pruning and show that common techniques of exclusive weight pruning increases the number of non-zero activations after ReLU."
337,Training Variational Auto Encoders with Discrete Latent Representations using Importance Sampling,"The Variational Auto Encoder is a popular generativelatent variable model that is oftenapplied for representation learning.Standard VAEs assume continuous valuedlatent variables and are trained by maximizationof the evidence lower bound.Conventional methods obtain adifferentiable estimate of the ELBO with reparametrized sampling andoptimize it with Stochastic Gradient Descend.However, this is not possible ifwe want to train VAEs with discrete valued latent variables,since reparametrized sampling is not possible.Till now, thereexist no simple solutions to circumvent this problem.In this paper, we propose an easy method to train VAEswith binary or categorically valued latent representations.Therefore, we use a differentiableestimator for the ELBO which is based on importance sampling.In experiments, we verify the approach andtrain two different VAEs architectures with Bernoulli andCategorically distributed latent representations on two different benchmarkdatasets.","We propose an easy method to train Variational Auto Encoders (VAE) with discrete latent representations, using importance samplingIntroducting an importance sampling distribution and using samples from distribution to compute importance-weighted estimate of the gradientThis paper proposes to use important sampling to optimize VAE with discrete latent variables."
338,DANA: Scalable Out-of-the-box Distributed ASGD Without Retuning,"Distributed computing can significantly reduce the training time of neural networks.Despite its potential, however, distributed training has not been widely adopted: scaling the training process is difficult, and existing SGD methods require substantial tuning of hyperparameters and learning schedules to achieve sufficient accuracy when increasing the number of workers.In practice, such tuning can be prohibitively expensive given the huge number of potential hyperparameter configurations and the effort required to test each one. We propose DANA, a novel approach that scales out-of-the-box to large clusters using the same hyperparameters and learning schedule optimized for training on a single worker, while maintaining similar final accuracy without additional overhead.DANA estimates the future value of model parameters by adapting Nesterov Accelerated Gradient to a distributed setting, and so mitigates the effect of gradient staleness, one of the main difficulties in scaling SGD to more workers.Evaluation on three state-of-the-art network architectures and three datasets shows that DANA scales as well as or better than existing work without having to tune any hyperparameters or tweak the learning schedule.For example, DANA achieves 75.73% accuracy on ImageNet when training ResNet-50 with 16 workers, similar to the non-distributed baseline.","A new distributed asynchronous SGD algorithm that achieves state-of-the-art accuracy on existing architectures without any additional tuning or overhead.Proposes an improvement to existing ASGD approaches at mid-size scaling using momentum with SGD for asynchronous training across a distributed worker pool.This paper addresses the gradient staleness vs parallel performance problem in distributed deep learning training, and proposes an approach to estimate future model parameters at the slaves to reduce communication latency effects."
339,Local Critic Training of Deep Neural Networks,"This paper proposes a novel approach to train deep neural networks by unlocking the layer-wise dependency of backpropagation training.The approach employs additional modules called local critic networks besides the main network model to be trained, which are used to obtain error gradients without complete feedforward and backward propagation processes.We propose a cascaded learning strategy for these local networks.In addition, the approach is also useful from multi-model perspectives, including structural optimization of neural networks, computationally efficient progressive inference, and ensemble classification for performance improvement.Experimental results show the effectiveness of the proposed approach and suggest guidelines for determining appropriate algorithm parameters.","We propose a new learning algorithm of deep neural networks, which unlocks the layer-wise dependency of backpropagation.An alternative training paradigm for DNIs in which the auxiliary module is trained to approximate directly the final output of the original model, offering side benefits.Describes a method of training neural networks without update locking."
340,Unbiasing Truncated Backpropagation Through Time," is a widespread method for learning recurrent computational graphs.Truncated BPTT keeps the computational benefits of while relieving the need for a complete backtrack through the whole data sequence at every step. However, truncation favors short-term dependencies: the gradient estimate of truncated BPTT is biased, so that it does not benefit from the convergence guarantees from stochastic gradient theory.We introduce , an algorithm that keeps the computational benefits of truncated BPTT, while providing unbiasedness.ARTBP works by using variable truncation lengths together with carefully chosen compensation factors in the backpropagation equation.We check the viability of ARTBP on two tasks.First, a simple synthetic task where careful balancing of temporal dependencies at different scales is needed: truncated BPTT displays unreliable performance, and in worst case scenarios, divergence, while ARTBP converges reliably.Second, on Penn Treebank character-level language modelling , ARTBP slightly outperforms truncated BPTT.",Provides an unbiased version of truncated backpropagation by sampling truncation lengths and reweighting accordingly.Proposes stochastic determination methods for truncation points in backpropagation through time.A new approximation to backpropagation through time to overcome the computational and memory loads that arise when having to learn from long sequences.
341,Transfer Learning for Sequences via Learning to Collocate,"Transfer learning aims to solve the data sparsity for a specific domain by applying information of another domain.Given a sequence, the transfer learning, usually enabled by recurrent neural network, represent the sequential information transfer.RNN uses a chain of repeating cells to model the sequence data.However, previous studies of neural network based transfer learning simply transfer the information across the whole layers, which are unfeasible for seq2seq and sequence labeling.Meanwhile, such layer-wise transfer learning mechanisms also lose the fine-grained cell-level information from the source domain.In this paper, we proposed the aligned recurrent transfer, ART, to achieve cell-level information transfer.ART is in a recurrent manner that different cells share the same parameters.Besides transferring the corresponding information at the same position, ART transfers information from all collocated words in the source domain.This strategy enables ART to capture the word collocation across domains in a more flexible way.We conducted extensive experiments on both sequence labeling tasks and sentence classification.ART outperforms the state-of-the-arts over all experiments.",Transfer learning for sequence via learning to align cell-level information across domains.The paper proposed to use RNN/LSTM with collocation alignment as a representation learning method for transfer learning/domain adaptation in NLP.
342,Bayesian Policy Optimization for Model Uncertainty,"Addressing uncertainty is critical for autonomous systems to robustly adapt to the real world.We formulate the problem of model uncertainty as a continuous Bayes-Adaptive Markov Decision Process, where an agent maintains a posterior distribution over latent model parameters given a history of observations and maximizes its expected long-term reward with respect to this belief distribution.Our algorithm, Bayesian Policy Optimization, builds on recent policy optimization algorithms to learn a universal policy that navigates the exploration-exploitation trade-off to maximize the Bayesian value function.To address challenges from discretizing the continuous latent parameter space, we propose a new policy network architecture that encodes the belief distribution independently from the observable state.Our method significantly outperforms algorithms that address model uncertainty without explicitly reasoning about belief distributions and is competitive with state-of-the-art Partially Observable Markov Decision Process solvers.","We formulate model uncertainty in Reinforcement Learning as a continuous Bayes-Adaptive Markov Decision Process and present a method for practical and scalable Bayesian policy optimization.Using a Bayesian approach, there is a better trade-off between exploration and exploitation in RL"
343,Towards GAN Benchmarks Which Require Generalization,"For many evaluation metrics commonly used as benchmarks for unconditional image generation, trivially memorizing the training set attains a better score than models which are considered state-of-the-art; we consider this problematic.We clarify a necessary condition for an evaluation metric not to behave this way: estimating the function must require a large sample from the model.In search of such a metric, we turn to neural network divergences, which are defined in terms of a neural network trained to distinguish between distributions.The resulting benchmarks cannot be won'' by training set memorization, while still being perceptually correlated and computable only from samples.We survey past work on using NNDs for evaluation, implement an example black-box metric based on these ideas, and validate experimentally that it can measure a notion of generalization.",We argue that GAN benchmarks must require a large sample from the model to penalize memorization and investigate whether neural network divergences have this property.Authors propose criterion for evaluating the quality of samples produced by a Generative Adversarial Network.
344,Towards Interpretable Chit-chat: Open Domain Dialogue Generation with Dialogue Acts,"Conventional methods model open domain dialogue generation as a black box through end-to-end learning from large scale conversation data.In this work, we make the first step to open the black box by introducing dialogue acts into open domain dialogue generation.The dialogue acts are generally designed and reveal how people engage in social chat.Inspired by analysis on real data, we propose jointly modeling dialogue act selection and response generation, and perform learning with human-human conversations tagged with a dialogue act classifier and a reinforcement approach to further optimizing the model for long-term conversation.With the dialogue acts, we not only achieve significant improvement over state-of-the-art methods on response quality for given contexts and long-term conversation in both machine-machine simulation and human-machine conversation, but also are capable of explaining why such achievements can be made.",open domain dialogue generation with dialogue actsThe authors use a distant supervision technique to add dialogue act tags as a conditioning factor for generating responses in open-domain dialoguesThe paper describes a technique to incorporate dialog acts into neural conversational agents
345,The Role of Minimal Complexity Functions in Unsupervised Learning of Semantic Mappings,"We discuss the feasibility of the following learning problem: given unmatched samples from two domains and nothing else, learn a mapping between the two, which preserves semantics.Due to the lack of paired samples and without any definition of the semantic information, the problem might seem ill-posed.Specifically, in typical cases, it seems possible to build infinitely many alternative mappings from every target mapping.This apparent ambiguity stands in sharp contrast to the recent empirical success in solving this problem.We identify the abstract notion of aligning two domains in a semantic way with concrete terms of minimal relative complexity.A theoretical framework for measuring the complexity of compositions of functions is developed in order to show that it is reasonable to expect the minimal complexity mapping to be unique.The measured complexity used is directly related to the depth of the neural networks being learned and a semantically aligned mapping could then be captured simply by learning using architectures that are not much bigger than the minimal architecture.Various predictions are made based on the hypothesis that semantic alignment can be captured by the minimal mapping.These are verified extensively.In addition, a new mapping algorithm is proposed and shown to lead to better mapping results.","Our hypothesis is that given two domains, the lowest complexity mapping that has a low discrepancy approximates the target mapping.The paper addresses the problem of learning mappings between different domains without any supervision, stating three conjectures.Demonstrates that in unsupervised learning on unaligned data it is possible to learn the between domains mapping using GAN only without a reconstruction loss."
346,Bridging HMMs and RNNs through Architectural Transformations,"A distinct commonality between HMMs and RNNs is that they both learn hidden representations for sequential data.In addition, it has been noted that the backward computation of the Baum-Welch algorithm for HMMs is a special case of the back-propagation algorithm used for neural networks).Do these observations suggest that, despite their many apparent differences, HMMs are a special case of RNNs?In this paper, we show that that is indeed the case, and investigate a series of architectural transformations between HMMs and RNNs, both through theoretical derivations and empirical hybridization.In particular, we investigate three key design factors—independence assumptions between the hidden states and the observation, the placement of softmaxes, and the use of non-linearities—in order to pin down their empirical effects.We present a comprehensive empirical study to provide insights into the interplay between expressivity and interpretability in this model family with respect to language modeling and parts-of-speech induction.","Are HMMs a special case of RNNs? We investigate a series of architectural transformations between HMMs and RNNs, both through theoretical derivations and empirical hybridization and provide new insights.This paper explores if HMMs are a special case of RNNs using language modeling and POS tagging"
347,Disentangled activations in deep networks,"Deep neural networks have been tremendously successful in a number of tasks.One of the main reasons for this is their capability to automaticallylearn representations of data in levels of abstraction,increasingly disentangling the data as the internal transformations are applied.In this paper we propose a novel regularization method that penalize covariance between dimensions of the hidden layers in a network, something that benefits the disentanglement.This makes the network learn nonlinear representations that are linearly uncorrelated, yet allows the model to obtain good results on a number of tasks, as demonstrated by our experimental evaluation.The proposed technique can be used to find the dimensionality of the underlying data, because it effectively disables dimensions that aren't needed.Our approach is simple and computationally cheap, as it can be applied as a regularizer to any gradient-based learning model.",We propose a novel regularization method that penalize covariance between dimensions of the hidden layers in a network.This paper presents a regularization mechanism which penalizes covariance between all dimensions in the latent representation of a neural network in order to disentangle the latent representation
348,Filter Training and Maximum Response: Classification via Discerning,"This report introduces a training and recognition scheme, in which classification is realized via class-wise discerning.Trained with datasets whose labels are randomly shuffled except for one class of interest, a neural network learns class-wise parameter values, and remolds itself from a feature sorter into feature filters, each of which discerns objects belonging to one of the classes only.Classification of an input can be inferred from the maximum response of the filters.A multiple check with multiple versions of filters can diminish fluctuation and yields better performance.This scheme of discerning, maximum response and multiple check is a method of general viability to improve performance of feedforward networks, and the filter training itself is a promising feature abstraction procedure.In contrast to the direct sorting, the scheme mimics the classification process mediated by a series of one component picking.","The proposed scheme mimics the classification process mediated by a series of one component picking.A method to increase accuracy of deep-nets on multi-class classification tasks seemingly by a reduction of multi-class to binary classification.A novel classification procedure of discerning, maxmum response, and multiple check to improve accuracy of mediocre networks and enhance feedforward networks."
349,Empirically Characterizing Overparameterization Impact on Convergence,"A long-held conventional wisdom states that larger models train more slowly when using gradient descent.This work challenges this widely-held belief, showing that larger models can potentially train faster despite the increasing computational requirements of each training step.In particular, we study the effect of network structure on halting time and show that larger models wider models in particular take fewer training steps to converge.We design simple experiments to quantitatively characterize the effect of overparametrization on weight space traversal.Results show that halting time improves when growing model's width for three different applications, and the improvement comes from each factor: The distance from initialized weights to converged weights shrinks with a power-law-like relationship, the average step size grows with a power-law-like relationship, and gradient vectors become more aligned with each other during traversal.","Empirically shows that larger models train in fewer training steps, because all factors in weight space traversal improve.This paper shows that wider RNNs improve convergence speed when applied to NLP problems, and by extension the effect of increasing the widths in deep neural networks on the convergence of optimizationThis paper characterizes the impact of over-parametrization in the number of iterations it takes an algorithm to converge, and presents further empirical observations on the effects of over-parametrization in neural network training."
350,Neural Program Repair by Jointly Learning to Localize and Repair,"Due to its potential to improve programmer productivity and software quality, automated program repair has been an active topic of research.Newer techniques harness neural networks to learn directly from examples of buggy programs and their fixes.In this work, we consider a recently identified class of bugs called variable-misuse bugs.The state-of-the-art solution for variable misuse enumerates potential fixes for all possible bug locations in a program, before selecting the best prediction.We show that it is beneficial to train a model that jointly and directly localizes and repairs variable-misuse bugs.We present multi-headed pointer networks for this purpose, with one head each for localization and repair.The experimental results show that the joint model significantly outperforms an enumerative solution that uses a pointer based model for repair alone.","Multi-headed Pointer Networks for jointly learning to localize and repair Variable Misuse bugsProposes a LSTM based model with pointers to break the problem of VarMisuse down into multiple steps.This paper presents an LSTM-based model for bug detection and repair of the VarMisuse bug, and demonstrates significant improvements compared to prior approaches on several datasets."
351,Human-like Clustering with Deep Convolutional Neural Networks,"Classification and clustering have been studied separately in machine learning and computer vision.Inspired by the recent success of deep learning models in solving various vision problems and the fact that humans serve as the gold standard in assessing clustering algorithms, here, we advocate for a unified treatment of the two problems and suggest that hierarchical frameworks that progressively build complex patterns on top of the simpler ones offer a promising solution.We do not dwell much on the learning mechanisms in these frameworks as they are still a matter of debate, with respect to biological constraints.Instead, we emphasize on the compositionality of the real world structures and objects.In particular, we show that CNNs, trained end to end using back propagation with noisy labels, are able to cluster data points belonging to several overlapping shapes, and do so much better than the state of the art algorithms.The main takeaway lesson from our study is that mechanisms of human vision, particularly the hierarchal organization of the visual ventral stream should be taken into account in clustering algorithms to reach human level clustering performance.This, by no means, suggests that other methods do not hold merits.For example, methods relying on pairwise affinities have been very successful in many cases but still fail in some cases.",Human-like Clustering with CNNsThe paper validates the idea that deep convolutional neural networks could learn to cluster input data better than other clustering methods by noting their ability to interpret the context of every input point due to a large field of view.This work combines deep learning for feature representation with the task of human-like unsupervised grouping.
352,L-Shapley and C-Shapley: Efficient Model Interpretation for Structured Data,"Instancewise feature scoring is a method for model interpretation, which yields, for each test instance, a vector of importance scores associated with features.Methods based on the Shapley score have been proposed as a fair way of computing feature attributions, but incur an exponential complexity in the number of features. This combinatorial explosion arises from the definition of Shapley value and prevents these methods from being scalable to large data sets and complex models.We focus on settings in which the data have a graph structure, and the contribution of features to the target variable is well-approximated by a graph-structured factorization. In such settings, we develop two algorithms with linear complexity for instancewise feature importance scoring on black-box models. We establish the relationship of our methods to the Shapley value and a closely related concept known as the Myerson value from cooperative game theory.We demonstrate on both language and image data that our algorithms compare favorably with other methods using both quantitative metrics and human evaluation.","We develop two linear-complexity algorithms for model-agnostic model interpretation based on the Shapley value, in the settings where the contribution of features to the target is well-approximated by a graph-structured factorization.The paper proposes two approximations to the Shapley value used for generating feature scores for interpretability.This paper proposes two methods for instance-wise feature importance scoring using Shapely values, and provides two efficient methods of computing approximate Shapely values when there is a known structure relating the features."
353,When and where do feed-forward neural networks learn localist representations?,"According to parallel distributed processing theory in psychology, neural networks learn distributed rather than interpretable localist representations.This view has been held so strongly that few researchers have analysed single units to determine if this assumption is correct.However, recent results from psychology, neuroscience and computer science have shown the occasional existence of local codes emerging in artificial and biological neural networks.In this paper, we undertake the first systematic survey of when local codes emerge in a feed-forward neural network, using generated input and output data with known qualities.We find that the number of local codes that emerge from a NN follows a well-defined distribution across the number of hidden layer neurons, with a peak determined by the size of input data, number of examples presented and the sparsity of input data.Using a 1-hot output code drastically decreases the number of local codes on the hidden layer.The number of emergent local codes increases with the percentage of dropout applied to the hidden layer, suggesting that the localist encoding may offer a resilience to noisy networks.This data suggests that localist coding can emerge from feed-forward PDP networks and suggests some of the conditions that may lead to interpretable localist representations in the cortex.The findings highlight how local codes should not be dismissed out of hand.","Local codes have been found in feed-forward neural networksA method for determining to what degree individual neurons in a hidden layer of an MLP encode a localist code, which is studied for different input representations.Studies the development of localist representations in the hidden layers of feed-forward neural networks."
354,Embedding Multimodal Relational Data,"Representing entities and relations in an embedding space is a well-studied approach for machine learning on relational data.Existing approaches however primarily focus on simple link structure between a finite set of entities, ignoring the variety of data types that are often used in relational databases, such as text, images, and numerical values.In our approach, we propose a multimodal embedding using different neural encoders for this variety of data, and combine with existing models to learn embeddings of the entities.We extend existing datasets to create two novel benchmarks, YAGO-10-plus and MovieLens-100k-plus, that contain additional relations such as textual descriptions and images of the original entities.We demonstrate that our model utilizes the additional information effectively to provide further gains in accuracy.Moreover, we test our learned multimodal embeddings by using them to predict missing multimodal attributes.","Extending relational modeling to support multimodal data using neural encoders.This paper proposes to perform link prediction in Knowledge Bases by supplementing the original entities with multimodal information, and presents a model able to encode all sorts of information when scoring triples.The paper is about incorporating information from different modalities into link prediction approaches"
355,Learning Sparse Structured Ensembles with SG-MCMC and Network Pruning,"An ensemble of neural networks is known to be more robust and accurate than an individual network, however usually with linearly-increased cost in both training and testing.In this work, we propose a two-stage method to learn Sparse Structured Ensembles for neural networks.In the first stage, we run SG-MCMC with group sparse priors to draw an ensemble of samples from the posterior distribution of network parameters.In the second stage, we apply weight-pruning to each sampled network and then perform retraining over the remained connections.In this way of learning SSEs with SG-MCMC and pruning, we not only achieve high prediction accuracy since SG-MCMC enhances exploration of the model-parameter space, but also reduce memory and computation cost significantly in both training and testing of NN ensembles.This is thoroughly evaluated in the experiments of learning SSE ensembles of both FNNs and LSTMs.For example, in LSTM based language modeling, we obtain 21% relative reduction in LM perplexity by learning a SSE of 4 large LSTM models, which has only 30% of model parameters and 70% of computations in total, as compared to the baseline large LSTM LM.To the best of our knowledge, this work represents the first methodology and empirical study of integrating SG-MCMC, group sparse prior and network pruning together for learning NN ensembles.","Propose a novel method by integrating SG-MCMC sampling, group sparse prior and network pruning to learn Sparse Structured Ensemble (SSE) with improved performance and significantly reduced cost than traditional methods. The authors propose a procedure to generate an ensemble of sparse structured modelsA new framework for training ensemble neural networks that uses SG-MCMC methods within deep learning, and then increases computational efficiency by group sparsity+pruning.This paper explores the use of FNN and LSTMs to make bayesian model averaging more computationally feasible and improve average model performance."
356,Meta-Learning Probabilistic Inference for Prediction,"This paper introduces a new framework for data efficient and versatile learning.Specifically:1) We develop ML-PIP, a general framework for Meta-Learning approximate Probabilistic Inference for Prediction.ML-PIP extends existing probabilistic interpretations of meta-learning to cover a broad class of methods.2) We introduce Versa, an instance of the framework employing a flexible and versatile amortization network that takes few-shot learning datasets as inputs, with arbitrary numbers of shots, and outputs a distribution over task-specific parameters in a single forward pass.Versa substitutes optimization at test time with forward passes through inference networks, amortizing the cost of inference and relieving the need for second derivatives during training.3) We evaluate Versa on benchmark datasets where the method sets new state-of-the-art results, and can handle arbitrary number of shots, and for classification, arbitrary numbers of classes at train and test time.The power of the approach is then demonstrated through a challenging few-shot ShapeNet view reconstruction task.","Novel framework for meta-learning that unifies and extends a broad class of existing few-shot learning methods. Achieves strong performance on few-shot learning benchmarks without requiring iterative test-time inference. This work tackles few-shot learning from a probabilistic inference viewpoint, achieving state-of-the-art despite simpler setup than many competitors"
357,Partially Mutual Exclusive Softmax for Positive and Unlabeled data,"In recent years, softmax together with its fast approximations has become the de-facto loss function for deep neural networks with multiclass predictions.However, softmax is used in many problems that do not fully fit the multiclass framework and where the softmax assumption of mutually exclusive outcomes can lead to biased results.This is often the case for applications such as language modeling, next event prediction and matrix factorization, where many of the potential outcomes are not mutually exclusive, but are more likely to be independent conditionally on the state.To this end, for the set of problems with positive and unlabeled data, we propose a relaxation of the original softmax formulation, where, given the observed state, each of the outcomes are conditionally independent but share a common set of negatives.Since we operate in a regime where explicit negatives are missing, we create an adversarially-trained model of negatives and derive a new negative sampling and weighting scheme which we denote as Cooperative Importance Sampling.We show empirically the advantages of our newly introduced negative sampling scheme by pluging it in the Word2Vec algorithm and benching it extensively against other negative sampling schemes on both language modeling and matrix factorization tasks and show large lifts in performance.","Defining a partially mutual exclusive softmax loss for postive data and implementing a cooperative based sampling schemeThis paper presents Cooperative Importance Sampling towards resolving the problem of the mutually exclusive assumption of traditional softmax being biased when negative samples are not explicitly definedThis paper proposes PMES methods to relax the exclusive outcome assumption in softmax loss, demonstrating empirical merit in improving word2vec type of embedding models."
358,A Teacher Student Network For Faster Video Classification,"Over the past few years, various tasks involving videos such as classification, description, summarization and question answering have received a lot of attention.Current models for these tasks compute an encoding of the video by treating it as a sequence of images and going over every image in the sequence, which becomes computationally expensive for longer videos.In this paper, we focus on the task of video classification and aim to reduce the computational cost by using the idea of distillation.Specifically, we propose a Teacher-Student network wherein the teacher looks at all the frames in the video but the student looks at only a small fraction of the frames in the video.The idea is to then train the student to minimize the difference between the final representation computed by the student and the teacher and/or the difference between the distributions predicted by the teacher and the student.This smaller student network which involves fewer computations but still learns to mimic the teacher can then be employed at inference time for video classification.We experiment with the YouTube-8M dataset and show that the proposed student network can reduce the inference time by upto 30% with a negligent drop in the performance.","Teacher-Student framework for efficient video classification using fewer frames The paper proposes an idea to distill from a full video classification model a small model that only receives smaller number of frames.The authors present a teacher-student network to solve video classification problem, proposing serial and parallel training algorithms aimed at reducing computational costs."
359,On Unifying Deep Generative Models,"Deep generative models have achieved impressive success in recent years.Generative Adversarial Networks and Variational Autoencoders, as powerful frameworks for deep generative model learning, have largely been considered as two distinct paradigms and received extensive independent studies respectively.This paper aims to establish formal connections between GANs and VAEs through a new formulation of them.We interpret sample generation in GANs as performing posterior inference, and show that GANs and VAEs involve minimizing KL divergences of respective posterior and inference distributions with opposite directions, extending the two learning phases of classic wake-sleep algorithm, respectively.The unified view provides a powerful tool to analyze a diverse set of existing model variants, and enables to transfer techniques across research lines in a principled way.For example, we apply the importance weighting method in VAE literatures for improved GAN learning, and enhance VAEs with an adversarial mechanism that leverages generated samples.Experiments show generality and effectiveness of the transfered techniques.",A unified statistical view of the broad class of deep generative models The paper develops a framework interpreting GAN algorithms as performing a form of variational inference on a generative model reconstructing an indicator variable of whether a sample is from the true of generative data distributions.
360,Pearl: Prototype lEArning via Rule Lists,"Deep neural networks have demonstrated promising prediction and classification performance on many healthcare applications.However, the interpretability of those models are often lacking.On the other hand, classical interpretable models such as rule lists or decision trees do not lead to the same level of accuracy as deep neural networks and can often be too complex to interpret.In this work, we present PEARL, Prototype lEArning via Rule Lists, which iteratively uses rule lists to guide a neural network to learn representative data prototypes.The resulting prototype neural network provides accurate prediction, and the prediction can be easily explained by prototype and its guiding rule lists.Thanks to the prediction power of neural networks, the rule lists from prototypes are more concise and hence provide better interpretability.On two real-world electronic healthcare records datasets, PEARL consistently outperforms all baselines across both datasets, especially achieving performance improvement over conventional rule learning by up to 28% and over prototype learning by up to 3%.Experimental results also show the resulting interpretation of PEARL is simpler than the standard rule learning.","a method combining rule list learning and prototype learning Presents a new interpretable prediction framework, which combines rule based learning, prototype learning, and NNs, that is particularly applicable to longitudinal data.This paper aims at tackling the lack of interpretability of deep learning models, and propose Prototype lEArning via Rule Lists (PEARL), which combines rule learning and prototype learning to achieve more accurate classification and makes the task of interpretability simpler."
361,Deli-Fisher GAN: Stable and Efficient Image Generation With Structured Latent Generative Space,"Generative Adversarial Networks are powerful tools for realistic image generation.However, a major drawback of GANs is that they are especially hard to train, often requiring large amounts of data and long training time.In this paper we propose the Deli-Fisher GAN, a GAN that generates photo-realistic images by enforcing structure on the latent generative space using similar approaches in .The structure of the latent space we consider in this paper is modeled as a mixture of Gaussians, whose parameters are learned in the training process.Furthermore, to improve stability and efficiency, we use the Fisher Integral Probability Metric as the divergence measure in our GAN model, instead of the Jensen-Shannon divergence.We show by experiments that the Deli-Fisher GAN performs better than DCGAN, WGAN, and the Fisher GAN as measured by inception score.","This paper proposes a new Generative Adversarial Network that is more stable, more efficient, and produces better images than those of status-quo This paper combines Fisher-GAN and Deli-GANThis paper combines Deli-GAN, which has a mixture prior distribution in latent space, and Fisher GAN, which uses Fisher IPM instead of JSD as an objective."
362,Sensor Transformation Attention Networks,"Recent work on encoder-decoder models for sequence-to-sequence mapping has shown that integrating both temporal and spatial attentional mechanisms into neural networks increases the performance of the system substantially.We report on a new modular network architecture that applies an attentional mechanism not on temporal and spatial regions of the input, but on sensor selection for multi-sensor setups.This network called the sensor transformation attention network is evaluated in scenarios which include the presence of natural noise or synthetic dynamic noise.We demonstrate how the attentional signal responds dynamically to changing noise levels and sensor-specific noise, leading to reduced word error rates on both audio and visual tasks using TIDIGITS and GRID; and also on CHiME-3, a multi-microphone real-world noisy dataset.The improvement grows as more channels are corrupted as demonstrated on the CHiME-3 dataset.Moreover, the proposed STAN architecture naturally introduces a number of advantages including ease of removing sensors from existing architectures, attentional interpretability, and increased robustness to a variety of noise environments.",We introduce a modular multi-sensor network architecture with an attentional mechanism that enables dynamic sensor selection on real-world noisy data from CHiME-3.A generic neural architecture able to learn the attention that must be payed to different input channels depending on the relative quality of each sensor with respect to the others. Considers the use of attention for sensor or channel selection with results on TIDIGITS and GRID showing a benefit of attention over concatentation of features.
363,PrivyNet: A Flexible Framework for Privacy-Preserving Deep Neural Network Training,"Massive data exist among user local platforms that usually cannot support deep neural network training due to computation and storage resource constraints.Cloud-based training schemes provide beneficial services but suffer from potential privacy risks due to excessive user data collection.To enable cloud-based DNN training while protecting the data privacy simultaneously, we propose to leverage the intermediate representations of the data, which is achieved by splitting the DNNs and deploying them separately onto local platforms and the cloud.The local neural network is used to generate the feature representations.To avoid local training and protect data privacy, the local NN is derived from pre-trained NNs.The cloud NN is then trained based on the extracted intermediate representations for the target learning task.We validate the idea of DNN splitting by characterizing the dependency of privacy loss and classification accuracy on the local NN topology for a convolutional NN based image classification task.Based on the characterization, we further propose PrivyNet to determine the local NN topology, which optimizes the accuracy of the target learning task under the constraints on privacy loss, local computation, and storage.The efficiency and effectiveness of PrivyNet are demonstrated with CIFAR-10 dataset.","To enable cloud-based DNN training while protecting the data privacy simultaneously, we propose to leverage the intermediate data representations, which is achieved by splitting the DNNs and deploying them separately onto local platforms and the cloud.This paper proposes a technique to privatize data by learning a feature representation that is difficult to use for image reconstruction, but helpful for image classification."
364,Real-valued (Medical) Time Series Generation with Recurrent Conditional GANs,"Generative Adversarial Networks have shown remarkable success as a framework for training models to produce realistic-looking data.In this work, we propose a Recurrent GAN and Recurrent Conditional GAN to produce realistic real-valued multi-dimensional time series, with an emphasis on their application to medical data.RGANs make use of recurrent neural networks in the generator and the discriminator.In the case of RCGANs, both of these RNNs are conditioned on auxiliary information.We demonstrate our models in a set of toy datasets, where we show visually and quantitatively that they can successfully generate realistic time-series.We also describe novel evaluation methods for GANs, where we generate a synthetic labelled training dataset, and evaluate on a real test set the performance of a model trained on the synthetic data, and vice-versa.We illustrate with these metrics that RCGANs can generate time-series data useful for supervised training, with only minor degradation in performance on real test data.This is demonstrated on digit classification from ‘serialised’ MNIST and by training an early warning system on a medical dataset of 17,000 patients from an intensive care unit.We further discuss and analyse the privacy concerns that may arise when using RCGANs to generate realistic synthetic medical time series data, and demonstrate results from differentially private training of the RCGAN.","Conditional recurrent GANs for real-valued medical sequences generation, showing novel evaluation approaches and an empirical privacy analysis.Proposes to use synthetic data generated by GANs as a replacement for personally identifiable data in training ML models for privacy-sensitive applicationsThe authors propose a novel recurrent GAN architecture that generates continuous domain sequences, and evaluate it on several synthetic tasks and an ICU timeseries data task.Proposes to use RGANs and RCGANs to generate synthetic sequences of actual data."
365,A Baseline Study of Emphasis Effects in Information Visualization,"Emphasis effects – visual changes that make certain elements moreprominent – are commonly used in information visualization to drawthe user’s attention or to indicate importance.Although theoreticalframeworks of emphasis exist, most metrics for predicting how emphasis effectswill be perceived by users come from abstract models of humanvision which may not apply to visualization design.In particular,it is difficult for designers to know, when designing a visualization,how different emphasis effects will compare and what level of oneeffect is equivalent to what level of another.To address this gap,we carried out two studies that provide empirical evidence abouthow users perceive different emphasis effects, using three visualvariables and eight strength levels.Results from gaze tracking, mouse clicks, and subjective responsesshow that there are significant differences between visual variablesand between levels, and allow us to develop an initial understandingof perceptual equivalence.We developed a model from the data inour first study, and used it to predict the results in the second; themodel was accurate, with high correlations between predictions andreal values.Our studies and empirical models provide valuable newinformation for designers who want to understand and control howemphasis effects will be perceived by users.","Our studies and empirical models provide valuable new information for designers who want to understand and control how emphasis effects will be perceived by usersThis paper considers which visual highlighting is perceived faster in data visualization and how different highlighting methods compare to each otherTwo studies on the efficacy of emphasis effects, one assessing levels of useful differences, and one more applied using actual different visualizations for a more ecologically valid investigation."
366,Finding ReMO (Related Memory Object): A Simple neural architecture for Text based Reasoning,"Memory Network based models have shown a remarkable progress on the task of relational reasoning.Recently, a simpler yet powerful neural network module called Relation Network has been introduced.Despite its architectural simplicity, the time complexity of relation network grows quadratically with data, hence limiting its application to tasks with a large-scaled memory.We introduce Related Memory Network, an end-to-end neural network architecture exploiting both memory network and relation network structures.We follow memory network's four components while each component operates similar to the relation network without taking a pair of objects.As a result, our model is as simple as RN but the computational complexity is reduced to linear time.It achieves the state-of-the-art results in jointly trained bAbI-10k story-based question answering and bAbI dialog dataset.","A simple reasoning architecture based on the memory network (MemNN) and relation network (RN), reducing the time complexity compared to the RN and achieving state-of-the-are result on bAbI story based QA and bAbI dialog.Introduces Related Memory Network (RMN), an improvement over Relationship Networks (RN)."
367,Coupled Ensembles of Neural Networks,"We investigate in this paper the architecture of deep convolutional networks.Building on existing state of the art models, we propose a reconfiguration of the model parameters into several parallel branches at the global network level, with each branch being a standalone CNN.We show that this arrangement is an efficient way to significantly reduce the number of parameters while at the same time improving the performance.The use of branches brings an additional form of regularization.In addition to splitting the parameters into parallel branches, we propose a tighter coupling of these branches by averaging their log-probabilities.The tighter coupling favours the learning of better representations, even at the level of the individual branches, as compared to when each branch is trained independently.We refer to this branched architecture as ""coupled ensembles"".The approach is very generic and can be applied with almost any neural network architecture.With coupled ensembles of DenseNet-BC and parameter budget of 25M, we obtain error rates of 2.92%, 15.68% and 1.50% respectively on CIFAR-10, CIFAR-100 and SVHN tasks.For the same parameter budget, DenseNet-BC has an error rate of 3.46%, 17.18%, and 1.8% respectively. With ensembles of coupled ensembles, of DenseNet-BC networks, with 50M total parameters, we obtain error rates of 2.72%, 15.13% and 1.42% respectively on these tasks.","We show that splitting a neural network into parallel branches improves performance and that proper coupling of the branches improves performance even further.The work proposes a reconfiguration of the existing state-of-the-art CNN model using a new branching architecture, with better performance.This paper shows parameter-saving benefits of coupled ensembling.Presents a deep network architecture which processes data using multiple parallel branches and combines the posterior from these branches to compute the final scores."
368,NICE: noise injection and clamping estimation for neural network quantization,"Convolutional Neural Networks are very popular in many fields including computer vision, speech recognition, natural language processing, to name a few.Though deep learning leads to groundbreaking performance in these domains, the networks used are very demanding computationally and are far from real-time even on a GPU, which is not power efficient and therefore does not suit low power systems such as mobile devices.To overcome this challenge, some solutions have been proposed for quantizing the weights and activations of these networks, which accelerate the runtime significantly.Yet, this acceleration comes at the cost of a larger error.The NICE method proposed in this work trains quantized neural networks by noise injection and a learned clamping, which improve the accuracy.This leads to state-of-the-art results on various regression and classification tasks, e.g., ImageNet classification with architectures such as ResNet-18/34/50 with low as 3-bit weights and 3 -bit activations.We implement the proposed solution on an FPGA to demonstrate its applicability for low power real-time applications.","Combine noise injection, gradual quantization and activation clamping learning to achieve state-of-the-art 3,4 and 5 bit quantizationProposes to inject noise during training and clamp parameter values in a layer as well as activation output in neural network quantization.A method for quantization of deep neural networks for classification and regression, using noise injection, clamping with learned maximum activations, and gradual block quantization to perform on-par or better than state-of-the-art methods."
369,Transferring Knowledge across Learning Processes,"In complex transfer learning scenarios new tasks might not be tightly linked to previous tasks.Approaches that transfer information contained only in the final parameters of a source model will therefore struggle.Instead, transfer learning at at higher level of abstraction is needed.We propose Leap, a framework that achieves this by transferring knowledge across learning processes.We associate each task with a manifold on which the training process travels from initialization to final parameters and construct a meta-learning objective that minimizes the expected length of this path.Our framework leverages only information obtained during training and can be computed on the fly at negligible cost.We demonstrate that our framework outperforms competing methods, both in meta-learning and transfer learning, on a set of computer vision tasks.Finally, we demonstrate that Leap can transfer knowledge across learning processes in demanding reinforcement learning environments that involve millions of gradient steps.","We propose Leap, a framework that transfers knowledge across learning processes by minimizing the expected distance the training process travels on a task's loss surface."", 'The article proposes a novel meta-learning objective aimed at outperforming state-of-the-art approaches when dealing with collections of tasks that exhibit substantial between-task diversity"
370,Incremental Learning through Deep Adaptation,"Given an existing trained neural network, it is often desirable to learn new capabilities without hindering performance of those already learned.Existing approaches either learn sub-optimal solutions, require joint training, or incur a substantial increment in the number of parameters for each added task, typically as many as the original network.We propose a method called Deep Adaptation Networks that constrains newly learned filters to be linear combinations of existing ones.DANs preserve performance on the original task, require a fraction of the number of parameters compared to standard fine-tuning procedures and converge in less cycles of training to a comparable or better level of performance.When coupled with standard network quantization techniques, we further reduce the parameter cost to around 3% of the original with negligible or no loss in accuracy.The learned architecture can be controlled to switch between various learned representations, enabling a single network to solve a task from multiple different domains.We conduct extensive experiments showing the effectiveness of our method on a range of image classification tasks and explore different aspects of its behavior.","An alternative to transfer learning that learns faster, requires much less parameters (3-13 %), usually achieves better results and precisely preserves performance on old tasks.Controller modules for increment learning on image classification datasets"
371,Hyperbolic Attention Networks,"Recent approaches have successfully demonstrated the benefits of learning the parameters of shallow networks in hyperbolic space.We extend this line of work by imposing hyperbolic geometry on the embeddings used to compute the ubiquitous attention mechanisms for different neural networks architectures.By only changing the geometry of embedding of object representations, we can use the embedding space more efficiently without increasing the number of parameters of the model.Mainly as the number of objects grows exponentially for any semantic distance from the query, hyperbolic geometry --as opposed to Euclidean geometry-- can encode those objects without having any interference.Our method shows improvements in generalization on neural machine translation on WMT'14, learning on graphs and visual question answering tasks while keeping the neural representations compact.","We propose to incorporate inductive biases and operations coming from hyperbolic geometry to improve the attention mechanism of the neural networks.This paper replaces the dot-product similarity used in attention mechanisms with the negative hyperbolic distance, and applies it to the existing Transformer model, graph attention networks, and Relation NetworksThe authors propose a novel approach to improve relational-attention by changing the matching and aggregation functions to use hyperbolic geometric. "
372,A dynamic game approach to training robust deep policies,"We present a method for evaluating the sensitivity of deep reinforcement learning policies.We also formulate a zero-sum dynamic game for designing robust deep reinforcement learning policies.Our approach mitigates the brittleness of policies when agents are trained in a simulated environment and are later exposed to the real world where it is hazardous to employ RL policies.This framework for training deep RL policies involve a zero-sum dynamic game against an adversarial agent, where the goal is to drive the system dynamics to a saddle region.Using a variant of the guided policy search algorithm, our agent learns to adopt robust policies that require less samples for learning the dynamics and performs better than the GPS algorithm.Without loss of generality, we demonstrate that deep RL policies trained in this fashion will be maximally robust to a worst"" possible adversarial disturbances.","This paper demonstrates how H-infinity control theory can help better design robust deep policies for robot motor taksProposes to incorporate elements of robust control into guided policy research in order to devise a method that is resilient to perturbations and model mismatch.The paper presents a method for evaluating the sensitivity and robustness of deep RL policies, and proposes a dynamic game approach for learning robust policies."
373,Robustness of Classifiers to Universal Perturbations: A Geometric Perspective,"Deep networks have recently been shown to be vulnerable to universal perturbations: there exist very small image-agnostic perturbations that cause most natural images to be misclassified by such classifiers.In this paper, we provide a quantitative analysis of the robustness of classifiers to universal perturbations, and draw a formal link between the robustness to universal perturbations, and the geometry of the decision boundary.Specifically, we establish theoretical bounds on the robustness of classifiers under two decision boundary models.We show in particular that the robustness of deep networks to universal perturbations is driven by a key property of their curvature: there exist shared directions along which the decision boundary of deep networks is systematically positively curved.Under such conditions, we prove the existence of small universal perturbations.Our analysis further provides a novel geometric method for computing universal perturbations, in addition to explaining their properties.",Analysis of vulnerability of classifiers to universal perturbations and relation to the curvature of the decision boundary.The paper provides an interesting analysis linking the geometry of classifier decision boundaries to small universal adversarial perturbations.This paper discusses universal perturbations - perturbations that can mislead a trained classifier if added to most of input data points.The paper develops models which attempt to explain the existence of universal perturbations which fool neural networks
374,Guiding Policies with Language via Meta-Learning,"Behavioral skills or policies for autonomous agents are conventionally learned from reward functions, via reinforcement learning, or from demonstrations, via imitation learning.However, both modes of task specification have their disadvantages: reward functions require manual engineering, while demonstrations require a human expert to be able to actually perform the task in order to generate the demonstration.Instruction following from natural language instructions provides an appealing alternative: in the same way that we can specify goals to other humans simply by speaking or writing, we would like to be able to specify tasks for our machines.However, a single instruction may be insufficient to fully communicate our intent or, even if it is, may be insufficient for an autonomous agent to actually understand how to perform the desired task.In this work, we propose an interactive formulation of the task specification problem, where iterative language corrections are provided to an autonomous agent, guiding it in acquiring the desired skill.Our proposed language-guided policy learning algorithm can integrate an instruction and a sequence of corrections to acquire new skills very quickly.In our experiments, we show that this method can enable a policy to follow instructions and corrections for simulated navigation and manipulation tasks, substantially outperforming direct, non-interactive instruction following.","We propose a meta-learning method for interactively correcting policies with natural language.This paper provides a meta learning framework that shows how to learn new tasks in an interactive setup. Each task is learned through a reinforcement learning setup, and then the task is being updated by observing new instructions.This paper teaches agents to complete tasks via natural language instructions in an iterative process."
375,Tinkering with black boxes: counterfactuals uncover modularity in generative models,"Deep generative models such as Generative Adversarial Networks andVariational Auto-Encoders are important tools to capture and investigatethe properties of complex empirical data.However, the complexity of their innerelements makes their functionment challenging to assess and modify.In thisrespect, these architectures behave as black box models.In order to betterunderstand the function of such networks, we analyze their modularity based onthe counterfactual manipulation of their internal variables.Our experiments on thegeneration of human faces with VAEs and GANs support that modularity betweenactivation maps distributed over channels of generator architectures is achievedto some degree, can be used to better understand how these systems operate and allow meaningful transformations of the generated images without further training.erate and edit the content of generated images.","We investigate the modularity of deep generative models.The paper provides a way to investigate the modular structure of the deep generative model, with the key concept to distribute over channels of generator architectures."
376,Seq2SQL: Generating Structured Queries From Natural Language Using Reinforcement Learning ,"Relational databases store a significant amount of the worlds data.However, accessing this data currently requires users to understand a query language such as SQL.We propose Seq2SQL, a deep neural network for translating natural language questions to corresponding SQL queries.Our model uses rewards from in the loop query execution over the database to learn a policy to generate the query, which contains unordered parts that are less suitable for optimization via cross entropy loss.Moreover, Seq2SQL leverages the structure of SQL to prune the space of generated queries and significantly simplify the generation problem.In addition to the model, we release WikiSQL, a dataset of 80654 hand-annotated examples of questions and SQL queries distributed across 24241 tables fromWikipedia that is an order of magnitude larger than comparable datasets.By applying policy based reinforcement learning with a query execution environment to WikiSQL, Seq2SQL outperforms a state-of-the-art semantic parser, improving execution accuracy from 35.9% to 59.4% and logical form accuracy from 23.4% to 48.3%.","We introduce Seq2SQL, which translates questions to SQL queries using rewards from online query execution, and WikiSQL, a SQL table/question/query dataset orders of magnitude larger than existing datasets.A new semantic parsing dataset which focuses on generating SQL from natural language using a reinforcement-learning based model"
377,Explainable Adversarial Learning: Implicit Generative Modeling of Random Noise during Training for Adversarial Robustness,"We introduce Explainable Adversarial Learning, ExL, an approach for training neural networks that are intrinsically robust to adversarial attacks.We find that the implicit generative modeling of random noise with the same loss function used during posterior maximization, improves a model's understanding of the data manifold furthering adversarial robustness."", ""We prove our approach's efficacy and provide a simplistic visualization tool for understanding adversarial data, using Principal Component Analysis.Our analysis reveals that adversarial robustness, in general, manifests in models with higher variance along the high-ranked principal components.We show that models learnt with our approach perform remarkably well against a wide-range of attacks.Furthermore, combining ExL with state-of-the-art adversarial training extends the robustness of a model, even beyond what it is adversarially trained for, in both white-box and black-box attack scenarios.","Noise modeling at the input during discriminative training improves adversarial robustness. Propose PCA based evaluation metric for adversarial robustnessThis paper proposes, ExL, an adversarial training method using multiplicate noise that is shown to be helpful in defending against blackbox attacks on three datasets.This paper includes multiplicative noise N in training data to achieve adversarial robustness, when training on both model parameters theta and on the noise itself."
378,Invariance and Inverse Stability under ReLU,"We flip the usual approach to study invariance and robustness of neural networks by considering the non-uniqueness and instability of the inverse mapping.We provide theoretical and numerical results on the inverse of ReLU-layers.First, we derive a necessary and sufficient condition on the existence of invariance that provides a geometric interpretation.Next, we move to robustness via analyzing local effects on the inverse.To conclude, we show how this reverse point of view not only provides insights into key effects, but also enables to view adversarial examples from different perspectives.","We analyze the invertibility of deep neural networks by studying preimages of ReLU-layers and the stability of the inverse.This paper studies the volume of preimage of a ReLU network’s activation at a certain layer, and it builds on the piecewise linearity of a ReLU network’s forward function. This paper presents an analysis of the inverse invariance of ReLU networks and provides upper bounds on singular values of a train network."
379,Strength in Numbers: Trading-off Robustness and Computation via Adversarially-Trained Ensembles,"While deep learning has led to remarkable results on a number of challenging problems, researchers have discovered a vulnerability of neural networks in adversarial settings, where small but carefully chosen perturbations to the input can make the models produce extremely inaccurate outputs.This makes these models particularly unsuitable for safety-critical application domains where robustness is extremely important.Recent work has shown that augmenting training with adversarially generated data provides some degree of robustness against test-time attacks.In this paper we investigate how this approach scales as we increase the computational budget given to the defender.We show that increasing the number of parameters in adversarially-trained models increases their robustness, and in particular that ensembling smaller models while adversarially training the entire ensemble as a single model is a more efficient way of spending said budget than simply using a larger single model.Crucially, we show that it is the adversarial training of the ensemble, rather than the ensembling of adversarially trained models, which provides robustness.","Adversarial training of ensembles provides robustness to adversarial examples beyond that observed in adversarially trained models and independently-trained ensembles thereof. Proposes to train an ensemble of models jointly, where at each time step, a set of examples that are adversarial for the ensemble itself is incorporated in the learning."
380,Routing Networks: Adaptive Selection of Non-Linear Functions for Multi-Task Learning,"Multi-task learning with neural networks leverages commonalities in tasks to improve performance, but often suffers from task interference which reduces the benefits of transfer.To address this issue we introduce the routing network paradigm, a novel neural network and training algorithm.A routing network is a kind of self-organizing neural network consisting of two components: a router and a set of one or more function blocks.A function block may be any neural network – for example a fully-connected or a convolutional layer.Given an input the router makes a routing decision, choosing a function block to apply and passing the output back to the router recursively, terminating when a fixed recursion depth is reached.In this way the routing network dynamically composes different function blocks for each input.We employ a collaborative multi-agent reinforcement learning approach to jointly train the router and function blocks.We evaluate our model against cross-stitch networks and shared-layer baselines on multi-task settings of the MNIST, mini-imagenet, and CIFAR-100 datasets.Our experiments demonstrate a significant improvement in accuracy, with sharper convergence.In addition, routing networks have nearly constant per-task training cost while cross-stitch networks scale linearly with the number of tasks.On CIFAR100 we obtain cross-stitch performance levels with an 85% average reduction in training time.","routing networks: a new kind of neural network which learns to adaptively route its input for multi-task learningThe paper suggests to use a modular network with a controller which makes decisions, at each time step, regarding the next nodule to apply.The paper presents a novel formulation for learning the optimal architecture of a neural network in a multi-task learning framework by using multi-agent reinforcement learning to find a policy, and shows improvement over hard-coded architectures with shared layers."
381,Learning Sparse Neural Networks through L_0 Regularization,"We propose a practical method for norm regularization for neural networks: pruning the network during training by encouraging weights to become exactly zero.Such regularization is interesting since it can greatly speed up training and inference, and it can improve generalization.AIC and BIC, well-known model selection criteria, are special cases of regularization.However, since the norm of weights is non-differentiable, we cannot incorporate it directly as a regularization term in the objective function.We propose a solution through the inclusion of a collection of non-negative stochastic gates, which collectively determine which weights to set to zero.We show that, somewhat surprisingly, for certain distributions over the gates, the expected regularized objective is differentiable with respect to the distribution parameters.We further propose the distribution for the gates, which is obtained by stretching'' a binary concrete distribution and then transforming its samples with a hard-sigmoid.The parameters of the distribution over the gates can then be jointly optimized with the original network parameters.As a result our method allows for straightforward and efficient learning of model structures with stochastic gradient descent and allows for conditional computation in a principled way.We perform various experiments to demonstrate the effectiveness of the resulting approach and regularizer.",We show how to optimize the expected L_0 norm of parametric models with gradient descent and introduce a new distribution that facilitates hard gating.The authors introduce a gradient-based approach to minimize an objective function with an L0 sparse penalty to help learn sparse neural networks
382,Attention-based Graph Neural Network for Semi-supervised Learning,"Recently popularized graph neural networks achieve the state-of-the-art accuracy on a number of standard benchmark datasets for graph-based semi-supervised learning, improving significantly over existing approaches.These architectures alternate between a propagation layer that aggregates the hidden states of the local neighborhood and a fully-connected layer.Perhaps surprisingly, we show that a linear model, that removes all the intermediate fully-connected layers, is still able to achieve a performance comparable to the state-of-the-art models.This significantly reduces the number of parameters, which is critical for semi-supervised learning where number of labeled examples are small.This in turn allows a room for designing more innovative propagation layers.Based on this insight, we propose a novel graph neural network that removes all the intermediate fully-connected layers, and replaces the propagation layers with attention mechanisms that respect the structure of the graph.The attention mechanism allows us to learn a dynamic and adaptive local summary of the neighborhood to achieve more accurate predictions.In a number of experiments on benchmark citation networks datasets, we demonstrate that our approach outperforms competing methods.By examining the attention weights among neighbors, we show that our model provides some interesting insights on how neighbors influence each other.","We propose a novel attention-based interpretable Graph Neural Network architecture which outperforms the current state-of-the-art Graph Neural Networks in standard benchmark datasetsThe authors propose two extensions of GCNs, by removing intermediate non-linearities from the GCN computation and adding an attention mechanism in the aggregation layer.The paper proposes a semi supervised learning algorithm for graph node classification with is inspired from Graph Neural Networks."
383,Learning Mixed-Curvature Representations in Product Spaces,"The quality of the representations achieved by embeddings is determined by how well the geometry of the embedding space matches the structure of the data.Euclidean space has been the workhorse for embeddings; recently hyperbolic and spherical spaces have gained popularity due to their ability to better embed new types of structured data such as hierarchical data but most data is not structured so uniformly.We address this problem by proposing learning embeddings in a product manifold combining multiple copies of these model spaces, providing a space of heterogeneous curvature suitable for a wide variety of structures.We introduce a heuristic to estimate the sectional curvature of graph data and directly determine an appropriate signature the number of component spaces and their dimensions of the product manifold.Empirically, we jointly learn the curvature and the embedding in the product space via Riemannian optimization.We discuss how to define and compute intrinsic quantities such as means a challenging notion for product manifolds and provably learnable optimization functions.On a range of datasets and reconstruction tasks, our product space embeddings outperform single Euclidean or hyperbolic spaces used in previous works, reducing distortion by 32.55% on a Facebook social network dataset.We learn word embeddings and find that a product of hyperbolic spaces in 50 dimensions consistently improves on baseline Euclidean and hyperbolic embeddings, by 2.6points in Spearman rank correlation on similarity tasksand 3.4 points on analogy accuracy.","Product manifold embedding spaces with heterogenous curvature yield improved representations compared to traditional embedding spaces for a variety of structures.Proposes a dimensionality reduction method that embeds data into a product manifold of spherical, Euclidean, and hyperbolic manifolds. The algorithm is based on matching the geodesic distances on the product manifold to graph distances."
384,Neural-Guided Deductive Search for Real-Time Program Synthesis from Examples,"Synthesizing user-intended programs from a small number of input-output exam-ples is a challenging problem with several important applications like spreadsheetmanipulation, data wrangling and code refactoring.Existing synthesis systemseither completely rely on deductive logic techniques that are extensively hand-engineered or on purely statistical models that need massive amounts of data, and ingeneral fail to provide real-time synthesis on challenging benchmarks.In this work,we propose Neural Guided Deductive Search, a hybrid synthesis techniquethat combines the best of both symbolic logic techniques and statistical models.Thus, it produces programs that satisfy the provided specifications by constructionand generalize well on unseen examples, similar to data-driven systems.Ourtechnique effectively utilizes the deductive search framework to reduce the learningproblem of the neural component to a simple supervised learning setup.Further,this allows us to both train on sparingly available real-world data and still leveragepowerful recurrent neural network encoders.We demonstrate the effectivenessof our method by evaluating on real-world customer scenarios by synthesizingaccurate programs with up to 12× speed-up compared to state-of-the-art systems.",We integrate symbolic (deductive) and statistical (neural-based) methods to enable real-time program synthesis with almost perfect generalization from 1 input-output example.The paper presents a branch-and-bound approach to learn good programs where an LSTM is used to predict which branches in the search tree should lead to good programsProposes system that synthesizes programs from a single example that generalize better than prior state-of-the-art
385,Avoiding degradation in deep feed-forward networks by phasing out skip-connections,"A widely observed phenomenon in deep learning is the degradation problem: increasingthe depth of a network leads to a decrease in performance on both test and training data.Novel architectures such as ResNets and Highway networks have addressed this issue by introducing various flavors of skip-connections or gating mechanisms.However, the degradation problem persists in the context of plain feed-forward networks.In this work we propose a simple method to address this issue.The proposed method poses the learning of weights in deep networks as a constrained optimization problem where the presence of skip-connections is penalized by Lagrange multipliers.This allows for skip-connections to be introduced during the early stages of training and subsequently phased out in a principled manner.We demonstrate the benefits of such an approach with experiments on MNIST, fashion-MNIST, CIFAR-10 and CIFAR-100 where the proposed method is shown to greatly decrease the degradation effect and is often competitive with ResNets.","Phasing out skip-connections in a principled manner avoids degradation in deep feed-forward networks.The authors present a new training strategy, VAN, for training very deep feed-forward networks without skip connectionsThe paper introduces an architecture that linearly interpolates between ResNets and vanilla deep nets without skip connections."
386,Sparse Regularized Deep Neural Networks For Efficient Embedded Learning,"Deep learning is becoming more widespread in its application due to its power in solving complex classification problems.However, deep learning models often require large memory and energy consumption, which may prevent them from being deployed effectively on embedded platforms, limiting their applications.This work addresses the problem by proposing methods for compressing the memory footprint of the models, including reducing the number of weights and the number of bits to store each weight.Beside, applying with sparsity-inducing regularization, our work focuses on speeding up stochastic variance reduced gradients optimization on non-convex problem.Our method that mini-batch SVRG with1 regularization on non-convex problem has faster and smoother convergence rates than SGD by using adaptive learning rates.Experimental evaluation of our approach uses MNIST and CIFAR-10 datasets on LeNet-300-100 and LeNet-5 models, showing our approach can reduce the memory requirements both in the convolutional and fully connected layers by up to 60 without affecting their test accuracy.",Compression of Deep neural networks deployed on embedded device. The authors present an l-1 regularized SVRG based training algorithm that is able to force many weights of the network to be 0.This work reduces memory requirements.
387,Learning with Mental Imagery,"In this paper, we propose deep convolutional generative adversarial networks that learn to produce a 'mental image' of the input image as internal representation of a certain category of input data distribution. "", ""This mental image is what the DCGAN 'imagines' that the input image might look like under ideal conditions. "", 'The mental image contains a version of the input that is iconic, without any peculiarities that do not contribute to the ideal representation of the input data distribution within a category.A DCGAN learns this association by training an encoder to capture salient features from the original image and a decoder to convert salient features into its associated mental image representation. Our new approach, which we refer to as a Mental Image DCGAN, learns features that are useful for recognizing entire classes of objects, and that this in turn has the benefit of helping single and zero shot recognition. We demonstrate our approach on object instance recognition and handwritten digit recognition tasks.","Object instance recognition with adversarial autoencoders was performed with a novel 'mental image' target that is canonical representation of the input image."", 'The paper proposes a method to learn features for object recognition that is invariant to various transformations of the object, most notably object pose.This paper investigated the task of few shot recognition via a generated “mental image” as intermediate representation given the input image."
388,AUTOMATA GUIDED HIERARCHICAL REINFORCEMENT LEARNING FOR ZERO-SHOT SKILL COMPOSITION,"An obstacle that prevents the wide adoption of reinforcement learning in control systems is its need for a large number of interactions with the environment in order to master a skill.The learned skill usually generalizes poorly across domains and re-training is often necessary when presented with a new task.We present a framework that combines techniques in with .The set of techniques we provide allows for the convenient specification of tasks with logical expressions, learns hierarchical policies with well-defined intrinsic rewards using any RL methods and is able to construct new skills from existing ones without additional learning.We evaluate the proposed methods in a simple grid world simulation as well as simulation on a Baxter robot.",Combine temporal logic with hierarchical reinforcement learning for skill compositionThe paper offers a strategy for constructing a product MDP out of an original MDP and the automaton associated with an LTL formula.Proposes to join temporal logic with hierarchical reinforcement learning to simplify skill composition.
389,N-Ary Quantization for CNN Model Compression and Inference Acceleration,"The tremendous memory and computational complexity of Convolutional Neural Networks prevents the inference deployment on resource-constrained systems.As a result, recent research focused on CNN optimization techniques, in particular quantization, which allows weights and activations of layers to be represented with just a few bits while achieving impressive prediction performance.However, aggressive quantization techniques still fail to achieve full-precision prediction performance on state-of-the-art CNN architectures on large-scale classification tasks.In this work we propose a method for weight and activation quantization that is scalable in terms of quantization levels and easy to compute while maintaining the performance close to full-precision CNNs.Our weight quantization scheme is based on trainable scaling factors and a nested-means clustering strategy which is robust to weight updates and therefore exhibits good convergence properties.The flexibility of nested-means clustering enables exploration of various n-ary weight representations with the potential of high parameter compression.For activations, we propose a linear quantization strategy that takes the statistical properties of batch normalization into account.We demonstrate the effectiveness of our approach using state-of-the-art models on ImageNet.",We propose a quantization scheme for weights and activations of deep neural networks. This reduces the memory footprint substantially and accelerates inference.CNN model compression aand inference acceleration using quantization.
390,Preferences Implicit in the State of the World,"Reinforcement learning agents optimize only the features specified in a reward function and are indifferent to anything left out inadvertently.This means that we must not only specify what to do, but also the much larger space of what not to do.It is easy to forget these preferences, since these preferences are already satisfied in our environment.This motivates our key insight: when a robot is deployed in an environment that humans act in, the state of the environment is already optimized for what humans want.We can therefore use this implicit preference information from the state to fill in the blanks.We develop an algorithm based on Maximum Causal Entropy IRL and use it to evaluate the idea in a suite of proof-of-concept environments designed to show its properties.We find that information from the initial state can be used to infer both side effects that should be avoided as well as preferences for how the environment should be organized.Our code can be found at https://github.com/HumanCompatibleAI/rlsp.","When a robot is deployed in an environment that humans have been acting in, the state of the environment is already optimized for what humans want, and we can use this to infer human preferences.The authors propose to augment the explicitly stated reward function of an RL agent with auxiliary rewards/costs inferred from the initial state and a model of the state dynamicsThis work proposes a way to infer the implicit information in the initial state using IRL and combine the inferred reward with a specified reward."
391,ProbGAN: Towards Probabilistic GAN with Theoretical Guarantees,"Probabilistic modelling is a principled framework to perform model aggregation, which has been a primary mechanism to combat mode collapse in the context of Generative Adversarial Networks.In this paper, we propose a novel probabilistic framework for GANs, ProbGAN, which iteratively learns a distribution over generators with a carefully crafted prior.Learning is efficiently triggered by a tailored stochastic gradient Hamiltonian Monte Carlo with a novel gradient approximation to perform Bayesian inference.Our theoretical analysis further reveals that our treatment is the first probabilistic framework that yields an equilibrium where generator distributions are faithful to the data distribution.Empirical evidence on synthetic high-dimensional multi-modal data and image databases demonstrates the superiority of our method over both start-of-the-art multi-generator GANs and other probabilistic treatment for GANs.",A novel probabilistic treatment for GAN with theoretical guarantee.This paper proposes a Bayesian GAN that has theoretical guarantees of convergence to the real distribution and put likelihoods over the generator and discriminator with logarithms proportional to the traditional GAN objective functions.
392,The Manifold Assumption and Defenses Against Adversarial Perturbations,"In the adversarial-perturbation problem of neural networks, an adversary starts with a neural network model and a point that classifies correctly, and applies a to to produce another point that classifies ."", "" In this paper, we propose taking into account produced by models when studying adversarial perturbations, where a natural measure of confidence'' is |F|_infty$"", "". Motivated by a thought experiment based on the manifold assumption, we propose a goodness property'' of models which states that .We give formalizations of this property and examine existing robust training objectives in view of them.Interestingly, we find that a recent objective by Madry et al. encourages training a model that satisfies well our formal version of the goodness property, but has a weak control of points that are wrong but with low confidence.However, if Madry et al.'s model is indeed a good solution to their objective, then good and bad points are now distinguishable and we can try to embed uncertain points back to the closest confident region to get correct predictions.We thus propose embedding objectives and algorithms, and perform an empirical study using this method.Our experimental results are encouraging: Madry et al.'s model wrapped with our embedding procedure achieves almost perfect success rate in defending against attacks that the base model fails on, while retaining good generalization behavior.",Defending against adversarial perturbations of neural networks from manifold assumption The manuscript proposes two objective functions based on the manifold assumption as defense mechanisms against adversarial examples.Defending against adversarial attacks based on the manifold assumption of natural data
393,Single Shot Neural Architecture Search Via Direct Sparse Optimization,"Recently Neural Architecture Search has aroused great interest in both academia and industry, however it remains challenging because of its huge and non-continuous search space.Instead of applying evolutionary algorithm or reinforcement learning as previous works, this paper proposes a Direct Sparse Optimization NAS method.In DSO-NAS, we provide a novel model pruning view to NAS problem.In specific, we start from a completely connected block, and then introduce scaling factors to scale the information flow between operations.Next, we impose sparse regularizations to prune useless connections in the architecture.Lastly, we derive an efficient and theoretically sound optimization method to solve it.Our method enjoys both advantages of differentiability and efficiency, therefore can be directly applied to large datasets like ImageNet.Particularly, On CIFAR-10 dataset, DSO-NAS achieves an average test error 2.84%, while on the ImageNet dataset DSO-NAS achieves 25.4% test error under 600M FLOPs with 8 GPUs in 18 hours.","single shot neural architecture search via direct sparse optimizationPresents an architecture search method where connections are removed with sparse regularization.This paper proposes Direct Sparse Optimization, which is a method to obtain neural architectures on specific problems, at a reasonable computational cost.This paper proposes a neural architecture search method based on a direct sparse optimization"
394,Model compression via distillation and quantization,"Deep neural networks continue to make significant advances, solving tasks from image classification to translation or reinforcement learning.One aspect of the field receiving considerable attention is efficiently executing deep models in resource-constrained environments, such as mobile or embedded devices.This paper focuses on this problem, and proposes two new compression methods, which jointly leverage weight quantization and distillation of larger teacher networks into smaller student networks.The first method we propose is called quantized distillation and leverages distillation during the training process, by incorporating distillation loss, expressed with respect to the teacher, into the training of a student network whose weights are quantized to a limited set of levels.The second method, differentiable quantization, optimizes the location of quantization points through stochastic gradient descent, to better fit the behavior of the teacher model. We validate both methods through experiments on convolutional and recurrent architectures.We show that quantized shallow students can reach similar accuracy levels to full-precision teacher models, while providing order of magnitude compression, and inference speedup that is linear in the depth reduction.In sum, our results enable DNNs for resource-constrained environments to leverage architecture and accuracy advances developed on more powerful devices.","Obtains state-of-the-art accuracy for quantized, shallow nets by leveraging distillation. Proposes small and low-cost models by combining distillation and quantization for vision and neural machine translation experimentsThis paper presents a framework of using the teacher model to help the compression for the deep learning model in the context of model compression."
395,Backplay: 'Man muss immer umkehren',"Model-free reinforcement learning requires a large number of trials to learn a good policy, especially in environments with sparse rewards.We explore a method to improve the sample efficiency when we have access to demonstrations.Our approach, Backplay, uses a single demonstration to construct a curriculum for a given task.Rather than starting each training episode in the environment's fixed initial state, we start the agent near the end of the demonstration and move the starting point backwards during the course of training until we reach the initial state.Our contributions are that we analytically characterize the types of environments where Backplay can improve training speed, demonstrate the effectiveness of Backplay both in large grid worlds and a complex four player zero-sum game, and show that Backplay compares favorably to other competitive methods known to improve sample efficiency.This includes reward shaping, behavioral cloning, and reverse curriculum generation.","Learn by working backwards from a single demonstration, even an inefficient one, and progressively have the agent do more of the solving itself.This paper presents a method for increasing the efficiency of sparse reward RL methods through a backward curriculum on expert demonstrations. The paper presents a strategy for solving sparse reward tasks with RL by sampling initial states from demonstrations."
396,Integrating Episodic Memory into a Reinforcement Learning Agent Using Reservoir Sampling,"Episodic memory is a psychology term which refers to the ability to recall specific events from the past.We suggest one advantage of this particular type of memory is the ability to easily assign credit to a specific state when remembered information is found to be useful.Inspired by this idea, and the increasing popularity of external memory mechanisms to handle long-term dependencies in deep learning systems, we propose a novel algorithm which uses a reservoir sampling procedure to maintain an external memory consisting of a fixed number of past states.The algorithm allows a deep reinforcement learning agent to learn online to preferentially remember those states which are found to be useful to recall later on.Critically this method allows for efficient online computation of gradient estimates with respect to the write process of the external memory.Thus unlike most prior mechanisms for external memory it is feasible to use in an online reinforcement learning setting.","External memory for online reinforcement learning based on estimating gradients over a novel reservoir sampling technique.The paper proposes a modified approach to RL, where an additional ""episodic memory"" is kept by the agent and use a ""query network"" that based on the current state."
397,Sparse Persistent RNNs: Squeezing Large Recurrent Networks On-Chip,"Recurrent Neural Networks are powerful tools for solving sequence-based problems, but their efficacy and execution time are dependent on the size of the network. Following recent work in simplifying these networks with model pruning and a novel mapping of work onto GPUs, we design an efficient implementation for sparse RNNs. We investigate several optimizations and tradeoffs: Lamport timestamps, wide memory loads, and a bank-aware weight layout. With these optimizations, we achieve speedups of over 6x over the next best algorithm for a hidden layer of size 2304, batch size of 4, and a density of 30%. Further, our technique allows for models of over 5x the size to fit on a GPU for a speedup of 2x, enabling larger networks to help advance the state-of-the-art. We perform case studies on NMT and speech recognition tasks in the appendix, accelerating their recurrent layers by up to 3x.","Combining network pruning and persistent kernels into a practical, fast, and accurate network implementation.This paper introduces sparse persistent RNNs, a mechanism to add pruning to the existing work of stashing RNN weights on a chip."
398,Viterbi-based Pruning for Sparse Matrix with Fixed and High Index Compression Ratio,"Weight pruning has proven to be an effective method in reducing the model size and computation cost while not sacrificing the model accuracy.Conventional sparse matrix formats, however, involve irregular index structures with large storage requirement and sequential reconstruction process, resulting in inefficient use of highly parallel computing resources.Hence, pruning is usually restricted to inference with a batch size of one, for which an efficient parallel matrix-vector multiplication method exists.In this paper, a new class of sparse matrix representation utilizing Viterbi algorithm that has a high, and more importantly, fixed index compression ratio regardless of the pruning rate, is proposed.In this approach, numerous sparse matrix candidates are first generated by the Viterbi encoder, and then the one that aims to minimize the model accuracy degradation is selected by the Viterbi algorithm.The model pruning process based on the proposed Viterbi encoder and Viterbi algorithm is highly parallelizable, and can be implemented efficiently in hardware to achieve low-energy, high-performance index decoding process.Compared with the existing magnitude-based pruning methods, index data storage requirement can be further compressed by 85.2% in MNIST and 83.9% in AlexNet while achieving similar pruning rate.Even compared with the relative index compression technique, our method can still reduce the index storage requirement by 52.7% in MNIST and 35.5% in AlexNet.","We present a new pruning method and sparse matrix format to enable high index compression ratio and parallel index decoding process.The authors use Viterbi encoding to dramatically compress the sparse matrix index of a pruned network, reducing one of the main memory overheads and speeding up inference in the parallel setting."
399,Hierarchical and Interpretable Skill Acquisition in Multi-task Reinforcement Learning,"Learning policies for complex tasks that require multiple different skills is a major challenge in reinforcement learning.It is also a requirement for its deployment in real-world scenarios.This paper proposes a novel framework for efficient multi-task reinforcement learning.Our framework trains agents to employ hierarchical policies that decide when to use a previously learned policy and when to learn a new skill.This enables agents to continually acquire new skills during different stages of training.Each learned task corresponds to a human language description.Because agents can only access previously learned skills through these descriptions, the agent can always provide a human-interpretable description of its choices.In order to help the agent learn the complex temporal dependencies necessary for the hierarchical policy, we provide it with a stochastic temporal grammar that modulates when to rely on previously learned skills and when to execute new skills.We validate our approach on Minecraft games designed to explicitly test the ability to reuse previously learned skills while simultaneously learning new skills.","A novel hierarchical policy network which can reuse previously learned skills alongside and as subcomponents of new skills by discovering the underlying relations between skills.This paper aims to learn hierarchical policies by using a recursive policy structure regulated by a stochastic temporal grammarThis paper proposes an approach to learning hierarchical policies in a lifelong learning context by stacking policies and then using an explicit ""switch"" policy."
400,Visualizing and Understanding the Semantics of Embedding Spaces via Algebraic Formulae,"Embeddings are a fundamental component of many modern machine learning and natural language processing models.Understanding them and visualizing them is essential for gathering insights about the information they capture and the behavior of the models.State of the art in analyzing embeddings consists in projecting them in two-dimensional planes without any interpretable semantics associated to the axes of the projection, which makes detailed analyses and comparison among multiple sets of embeddings challenging.In this work, we propose to use explicit axes defined as algebraic formulae over embeddings to project them into a lower dimensional, but semantically meaningful subspace, as a simple yet effective analysis and visualization methodology.This methodology assigns an interpretable semantics to the measures of variability and the axes of visualizations, allowing for both comparisons among different sets of embeddings and fine-grained inspection of the embedding spaces.We demonstrate the power of the proposed methodology through a series of case studies that make use of visualizations constructed around the underlying methodology and through a user study.The results show how the methodology is effective at providing more profound insights than classical projection methods and how it is widely applicable to many other use cases.",We propose to use explicit vector algebraic formulae projection as an alternative way to visualize embedding spaces specifically tailored for goal-oriented analysis tasks and it outperforms t-SNE in our user study.Analysis of embedding psaces in a non-parametric (example-based_ way
401,Towards Resisting Large Data Variations via Introspective Learning,"Learning deep networks which can resist large variations between training andtesting data is essential to build accurate and robust image classifiers. Towardsthis end, a typical strategy is to apply data augmentation to enlarge the trainingset. However, standard data augmentation is essentially a brute-force strategywhich is inefficient, as it performs all the pre-defined transformations to everytraining sample.In this paper, we propose a principled approach to train networkswith significantly improved resistance to large variations between training andtesting data. This is achieved by embedding a learnable transformation moduleinto the introspective networks, which is a convolutional neural network classifier empowered withgenerative capabilities. Our approach alternatively synthesizes pseudo-negativesamples with learned transformations and enhances the classifier by retraining itwith synthesized samples. Experimental results verify that our approach signif-icantly improves the ability of deep networks to resist large variations betweentraining and testing data and achieves classification accuracy improvements onseveral benchmark datasets, including MNIST, affNIST, SVHN and CIFAR-10.","We propose a principled approach that endows classifiers with the ability to resist larger variations between training and testing data in an intelligent and efficient manner.Using introspective learning to handle data variations at test timeThis paper suggests the use of learned transformation networks, embedded within introspective networks to improve classification performance with synthesized examples."
402,Some Considerations on Learning to Explore via Meta-Reinforcement Learning,We consider the problem of exploration in meta reinforcement learning.Two new meta reinforcement learning algorithms are suggested: E-MAML and ERL2.Results are presented on a novel environment we call 'Krazy World' and a set of maze environments.We show E-MAML and ERL2 deliver better performance on tasks where exploration is important.,Modifications to MAML and RL2 that should allow for better exploration. The paper proposes a trick of extending objective functions to drive exploration in meta-RL on top of two recent meta-RL algorithms
403,Probabilistic Neural-Symbolic Models for Interpretable Visual Question Answering,"We propose a new class of probabilistic neural-symbolic models for visual question answering that provide interpretable explanations of their decision making in the form of programs, given a small annotated set of human programs.The key idea of our approach is to learn a rich latent space which effectively propagates program annotations from known questions to novel questions.We do this by formalizing prior work on VQA, called module networks as discrete, structured, latent variable models on the joint distribution over questions and answers given images, and devise a procedure to train the model effectively.Our results on a dataset of compositional questions about SHAPES show that our model generates more interpretable programs and obtains better accuracy on VQA in the low-data regime than prior work.","A probabilistic neural symbolic model with a latent program space, for more interpretable question answeringThis paper proposes a discrete, structured latent variable model for visual question answering that involves compositional generalization and reasoning with significant gain in performance and capability."
404,Ground-Truth Adversarial Examples,"The ability to deploy neural networks in real-world, safety-critical systems is severely limited by the presence of adversarial examples: slightly perturbed inputs that are misclassified by the network.In recent years, several techniques have been proposed for training networks that are robust to such examples; and each time stronger attacks have been devised, demonstrating the shortcomings of existing defenses.This highlights a key difficulty in designing an effective defense: the inability to assess a network's robustness against future attacks.We propose to address this difficulty through formal verification techniques.We construct ground truths: adversarial examples with a provably-minimal distance from a given input point.We demonstrate how ground truths can serve to assess the effectiveness of attack techniques, by comparing the adversarial examples produced by those attacks to the ground truths; and also of defense techniques, by computing the distance to the ground truths before and after the defense is applied, and measuring the improvement.We use this technique to assess recently suggested attack and defense techniques.",We use formal verification to assess the effectiveness of techniques for finding adversarial examples or for defending against adversarial examples.This paper proposes a method for computing adversarial examples with minimum distance to the original inputs.The authors propose to employ provably minimal-distance examples as a tool to evaluate the robustness of a trained network.The paper describes a method for generating adversarial examples that have minimal distance to the training example used to generate them
405,Multi-step Retriever-Reader Interaction for Scalable Open-domain Question Answering,"This paper introduces a new framework for open-domain question answering in which the retriever and the reader with each other.The framework is agnostic to the architecture of the machine reading model provided it has to the token-level hidden representations of the reader.The retriever uses fast nearest neighbor search that allows it to scale to corpora containing millions of paragraphs.A gated recurrent unit updates the query at each step conditioned on the of the reader and the query is used to re-rank the paragraphs by the retriever.We conduct analysis and show that iterative interaction helps in retrieving informative paragraphs from the corpus.Finally, we show that our multi-step-reasoning framework brings consistent improvement when applied to two widely used reader architectures on various large open-domain datasets \qau, quasart, searchqa, and squadofootnote}.","Paragraph retriever and machine reader interacts with each other via reinforcement learning to yield large improvements on open domain datasets The paper introduces a new framework of bi-directional interaction between document retriever and reader for open-domain question answering with idea of 'reader state' from reader to retriever."", 'The paper proposes a multi-document extractive machine reading model composed of 3 distinct parts and an algorithm."
406,Stacked U-Nets: A No-Frills Approach to Natural Image Segmentation,"Many imaging tasks require global information about all pixels in an image.Conventional bottom-up classification networks globalize information by decreasing resolution; features are pooled and down-sampled into a single output.But for semantic segmentation and object detection tasks, a network must provide higher-resolution pixel-level outputs.To globalize information while preserving resolution, many researchers propose the inclusion of sophisticated auxiliary blocks, but these come at the cost of a considerable increase in network size and computational cost.This paper proposes stacked u-nets, which iteratively combine features from different resolution scales while maintaining resolution.SUNets leverage the information globalization power of u-nets in a deeper net- work architectures that is capable of handling the complexity of natural images.SUNets perform extremely well on semantic segmentation tasks using a small number of parameters.","Presents new architecture which leverages information globalization power of u-nets in a deeper networks and performs well across tasks without any bells and whistles.A network architecture for semantic image segmentation, based on composing a stack of basic U-Net architectures, that reduces the number of parameters and improves results.This proposes a stacked U-Net architecture for image segmentation."
407,Topic-Based Question Generation,"Asking questions is an important ability for a chatbot.This paper focuses on question generation.Although there are existing works on question generation based on a piece of descriptive text, it remains to be a very challenging problem.In the paper, we propose a new question generation problem, which also requires the input of a target topic in addition to a piece of descriptive text.The key reason for proposing the new problem is that in practical applications, we found that useful questions need to be targeted toward some relevant topics.One almost never asks a random question in a conversation.Due to the fact that given a descriptive text, it is often possible to ask many types of questions, generating a question without knowing what it is about is of limited use.To solve the problem, we propose a novel neural network that is able to generate topic-specific questions.One major advantage of this model is that it can be trained directly using a question-answering corpus without requiring any additional annotations like annotating topics in the questions or answers.Experimental results show that our model outperforms the state-of-the-art baseline.",We propose a neural network that is able to generate topic-specific questions.Presents a neural network-based approach to generate topic-specific questions with the motivation that topical questions are more meaningful in practical applications.Proposes a topic-based generation method using an LSTM to extract topics using a two-stage encoding technique
408,Adversarial Domain Adaptation for Stable Brain-Machine Interfaces,"Brain-Machine Interfaces have recently emerged as a clinically viable optionto restore voluntary movements after paralysis.These devices are based on theability to extract information about movement intent from neural signals recordedusing multi-electrode arrays chronically implanted in the motor cortices of thebrain.However, the inherent loss and turnover of recorded neurons requires repeatedrecalibrations of the interface, which can potentially alter the day-to-dayuser experience.The resulting need for continued user adaptation interferes withthe natural, subconscious use of the BMI.Here, we introduce a new computationalapproach that decodes movement intent from a low-dimensional latent representationof the neural data.We implement various domain adaptation methodsto stabilize the interface over significantly long times.This includes CanonicalCorrelation Analysis used to align the latent variables across days; this methodrequires prior point-to-point correspondence of the time series across domains.Alternatively, we match the empirical probability distributions of the latent variablesacross days through the minimization of their Kullback-Leibler divergence.These two methods provide a significant and comparable improvement in the performanceof the interface.However, implementation of an Adversarial DomainAdaptation Network trained to match the empirical probability distribution of theresiduals of the reconstructed neural signals outperforms the two methods basedon latent variables, while requiring remarkably few data points to solve the domainadaptation problem.",We implement an adversarial domain adaptation network to stabilize a fixed Brain-Machine Interface against gradual changes in the recorded neural signals.Describes a novel approach for implanted brain-machine interface in order to address calibration problem and covariate shift. The authors define a BMI that uses an autoencoder and then address the problem of data drift in BMI.
409,Understanding Grounded Language Learning Agents,"Neural network-based systems can now learn to locate the referents of words and phrases in images, answer questions about visual scenes, and even execute symbolic instructions as first-person actors in partially-observable worlds.To achieve this so-called grounded language learning, models must overcome certain well-studied learning challenges that are also fundamental to infants learning their first words.While it is notable that models with no meaningful prior knowledge overcome these learning obstacles, AI researchers and practitioners currently lack a clear understanding of exactly how they do so.Here we address this question as a way of achieving a clearer general understanding of grounded language learning, both to inform future research and to improve confidence in model predictions.For maximum control and generality, we focus on a simple neural network-based language learning agent trained via policy-gradient methods to interpret synthetic linguistic instructions in a simulated 3D world.We apply experimental paradigms from developmental psychology to this agent, exploring the conditions under which established human biases and learning effects emerge.We further propose a novel way to visualise and analyse semantic representation in grounded language learning agents that yields a plausible computational account of the observed effects.",Analysing and understanding how neural network agents learn to understand simple grounded languageThe authors connect psychological experimental methods to understanding how the black box of deep learning methods solves problems.This paper presents an analysis of the agents who learn grounded language through reinforcement learning in a simple environment that combines verbal instruction with visual information
410,"Unsupervised Discovery of Parts, Structure, and Dynamics","Humans easily recognize object parts and their hierarchical structure by watching how they move; they can then predict how each part moves in the future.In this paper, we propose a novel formulation that simultaneously learns a hierarchical, disentangled object representation and a dynamics model for object parts from unlabeled videos.Our Parts, Structure, and Dynamics model learns to, first, recognize the object parts via a layered image representation; second, predict hierarchy via a structural descriptor that composes low-level concepts into a hierarchical structure; and third, model the system dynamics by predicting the future.Experiments on multiple real and synthetic datasets demonstrate that our PSD model works well on all three tasks: segmenting object parts, building their hierarchical structure, and capturing their motion distributions.","Learning object parts, hierarchical structure, and dynamics by watching how they moveProposes an unsupervised learning model that learns to disentangle objects into parts, predict hierarchical structure for the parts, and based on the disentangled parts and the hierarchy, predict motion."
411,Understanding Opportunities for Efficiency in Single-image Super Resolution Networks,"A successful application of convolutional architectures is to increase the resolution of single low-resolution images -- a image restoration task called super-resolution.Naturally, SR is of value to resource constrained devices like mobile phones, electronic photograph frames and televisions to enhance image quality.However, SR demands perhaps the most extreme amounts of memory and compute operations of any mainstream vision task known today, preventing SR from being deployed to devices that require them.In this paper, we perform a early systematic study of system resource efficiency for SR, within the context of a variety of architectural and low-precision approaches originally developed for discriminative neural networks.We present a rich set of insights, representative SR architectures, and efficiency trade-offs; for example, the prioritization of ways to compress models to reach a specific memory and computation target and techniques to compact SR models so that they are suitable for DSPs and FPGAs.As a result of doing so, we manage to achieve better and comparable performance with previous models in the existing literature, highlighting the practicality of using existing efficiency techniques in SR tasks.Collectively, we believe these results provides the foundation for further research into the little explored area of resource efficiency for SR.",We build an understanding of resource-efficient techniques on Super-ResolutionThe paper proposes a detailed empirical evaluation of the trade-offs achieved by various convolutional neural networks on the super resolution problem.This paper proposed to improve the system resource efficiency for super resolution networks.
412,On the Spectral Bias of Neural Networks,"Neural networks are known to be a class of highly expressive functions able to fit even random input-output mappings with 100% accuracy.In this work we present properties of neural networks that complement this aspect of expressivity.By using tools from Fourier analysis, we show that deep ReLU networks are biased towards low frequency functions, meaning that they cannot have local fluctuations without affecting their global behavior.Intuitively, this property is in line with the observation that over-parameterized networks find simple patterns that generalize across data samples.We also investigate how the shape of the data manifold affects expressivity by showing evidence that learning high frequencies gets easier with increasing manifold complexity, and present a theoretical understanding of this behavior.Finally, we study the robustness of the frequency components with respect to parameter perturbation, to develop the intuition that the parameters must be finely tuned to express high frequency functions.","We investigate ReLU networks in the Fourier domain and demonstrate peculiar behaviour.Fourier analysis of ReLU network, finding that they are biased towards learning low frequency This paper has theoretical and empirical contributions on topic of Fourier coefficients of neural networks"
413,Modeling Uncertainty with Hedged Instance Embeddings,"Instance embeddings are an efficient and versatile image representation that facilitates applications like recognition, verification, retrieval, and clustering.Many metric learning methods represent the input as a single point in the embedding space.Often the distance between points is used as a proxy for match confidence.However, this can fail to represent uncertainty which can arise when the input is ambiguous, e.g., due to occlusion or blurriness.This work addresses this issue and explicitly models the uncertainty by “hedging” the location of each input in the embedding space.We introduce the hedged instance embedding in which embeddings are modeled as random variables and the model is trained under the variational information bottleneck principle.Empirical results on our new N-digit MNIST dataset show that our method leads to the desired behavior of “hedging its bets” across the embedding space upon encountering ambiguous inputs.This results in improved performance for image matching and classification tasks, more structure in the learned embedding space, and an ability to compute a per-exemplar uncertainty measure which is correlated with downstream performance.",The paper proposes using probability distributions instead of points for instance embeddings tasks such as recognition and verification.Paper proposes an alternative to current point embedding and a technique to train them.The paper proposes a model using uncertain-embeddings to extend deep learning to Bayesian applications
414,Tensor Contraction & Regression Networks,"Convolution neural networks typically consist of many convolutional layers followed by several fully-connected layers. While convolutional layers map between high-order activation tensors, the fully-connected layers operate on flattened activation vectors. Despite its success, this approach has notable drawbacks.Flattening discards the multi-dimensional structure of the activations, and the fully-connected layers require a large number of parameters.We present two new techniques to address these problems. First, we introduce tensor contraction layers which can replace the ordinary fully-connected layers in a neural network.Second, we introduce tensor regression layers, which express the output of a neural network as a low-rank multi-linear mapping from a high-order activation tensor to the softmax layer. Both the contraction and regression weights are learned end-to-end by backpropagation.By imposing low rank on both, we use significantly fewer parameters. Experiments on the ImageNet dataset show that applied to the popular VGG and ResNet architectures, our methods significantly reduce the number of parameters in the fully connected layers while negligibly impacting accuracy.","We propose tensor contraction and low-rank tensor regression layers to preserve and leverage the multi-linear structure throughout the network, resulting in huge space savings with little to no impact on performance.This paper proposes new layer architectures of neural networks using a low-rank representation of tensorsThis paper incorporates tensor decomposition and tensor regression into CNN by using a new tensor regression layer."
415,INCORPORATING BILINGUAL DICTIONARIES FOR LOW RESOURCE SEMI-SUPERVISED NEURAL MACHINE TRANSLATION,"We explore ways of incorporating bilingual dictionaries to enable semi-supervisedneural machine translation.Conventional back-translation methods have shownsuccess in leveraging target side monolingual data.However, since the quality ofback-translation models is tied to the size of the available parallel corpora, thiscould adversely impact the synthetically generated sentences in a low resourcesetting.We propose a simple data augmentation technique to address both thisshortcoming.We incorporate widely available bilingual dictionaries that yieldword-by-word translations to generate synthetic sentences.This automaticallyexpands the vocabulary of the model while maintaining high quality content.Ourmethod shows an appreciable improvement in performance over strong baselines.",We use bilingual dictionaries for data augmentation for neural machine translationThis paper investigates using bilingual dictionaries to create synthetic sources for target-side monolingual data to improve over NMT models trained with small amounts of parallel data.
416,Episodic Curiosity through Reachability,"Rewards are sparse in the real world and most of today's reinforcement learning algorithms struggle with such sparsity.One solution to this problem is to allow the agent to create rewards for itself - thus making rewards dense and more suitable for learning.In particular, inspired by curious behaviour in animals, observing something novel could be rewarded with a bonus.Such bonus is summed up with the real task reward - making it possible for RL algorithms to learn from the combined reward.We propose a new curiosity method which uses episodic memory to form the novelty bonus.To determine the bonus, the current observation is compared with the observations in memory.Crucially, the comparison is done based on how many environment steps it takes to reach the current observation from those in memory - which incorporates rich information about environment dynamics.This allows us to overcome the known ""couch-potato"" issues of prior work - when the agent finds a way to instantly gratify itself by exploiting actions which lead to hardly predictable consequences.We test our approach in visually rich 3D environments in ViZDoom, DMLab and MuJoCo.In navigational tasks from ViZDoom and DMLab, our agent outperforms the state-of-the-art curiosity method ICM.In MuJoCo, an ant equipped with our curiosity module learns locomotion out of the first-person-view curiosity only.The code is available at https://github.com/google-research/episodic-curiosity/.","We propose a novel model of curiosity based on episodic memory and the ideas of reachability which allows us to overcome the known ""couch-potato"" issues of prior work.Proposes to give exploration bonuses in RL algorithms by giving larger bonuses to observations that are father away in environment steps.The authors propose an exploration bonus that is aimed to aid in sparse reward RL problems and considers many experiments on complex 3D environments"
417,Can Neural Networks Understand Logical Entailment?,"We introduce a new dataset of logical entailments for the purpose of measuring models' ability to capture and exploit the structure of logical expressions against an entailment prediction task."", ""We use this task to compare a series of architectures which are ubiquitous in the sequence-processing literature, in addition to a new model class PossibleWorldNets which computes entailment as a convolution over possible worlds''.Results show that convolutional networks present the wrong inductive bias for this class of problems relative to LSTM RNNs, tree-structured neural networks outperform LSTM RNNs due to their enhanced ability to exploit the syntax of logic, and PossibleWorldNets outperform all benchmarks.","We introduce a new dataset of logical entailments for the purpose of measuring models' ability to capture and exploit the structure of logical expressions against an entailment prediction task."", 'The paper proposes a new model to use deep models for detecting logical entailment as a product of continuous functions over possible worlds.Proposes a new model designed for machine learning with predicting logical entailment."
418,Incremental Learning in Deep Convolutional Neural Networks Using Partial Network Sharing,"Deep convolutional neural network based supervised learning is a widely practiced approach for large-scale image classification. However, retraining these large networks to accommodate new, previously unseen data demands high computational time and energy requirements.Also, previously seen training samples may not be available at the time of retraining.We propose an efficient training methodology and incrementally growing a DCNN to allow new classes to be learned while sharing part of the base network.Our proposed methodology is inspired by transfer learning techniques, although it does not forget previously learned classes.An updated network for learning new set of classes is formed using previously learned convolutional layers with addition of few newly added convolutional kernels included in the later layers of the network.We evaluated the proposed scheme on several recognition applications.The classification accuracy achieved by our approach is comparable to the regular incremental learning approach.","The paper is about a new energy-efficient methodology for Incremental learningProposes procedure for incremental learning as transfer learning.The paper presents a method to train deep convolutional neural networks in an incremental fashion, in which data are available in small batches over a period of time.Presents an approach to class-incremental learning using deep networks by proposing three different learning strategies in the final/best approach."
419,Parallelizing Linear Recurrent Neural Nets Over Sequence Length,"Recurrent neural networks are widely used to model sequential data buttheir non-linear dependencies between sequence elements prevent parallelizingtraining over sequence length.We show the training of RNNs with only linearsequential dependencies can be parallelized over the sequence length using theparallel scan algorithm, leading to rapid training on long sequences even withsmall minibatch size.We develop a parallel linear recurrence CUDA kernel andshow that it can be applied to immediately speed up training and inference ofseveral state of the art RNN architectures by up to 9x. We abstract recent workon linear RNNs into a new framework of linear surrogate RNNs and develop alinear surrogate model for the long short-term memory unit, the GILR-LSTM, thatutilizes parallel linear recurrence. We extend sequence learning to newextremely long sequence regimes that were previously out of reach bysuccessfully training a GILR-LSTM on a synthetic sequence classification taskwith a one million timestep dependency.","use parallel scan to parallelize linear recurrent neural nets. train model on length 1 million dependencyProposes accelerating RNN by applying the method from Blelloch.The authors propose a parallel algorithm for Linear Surrogate RNNs, which produces speedups over the existing implements of Quasi-RNN, SRU, and LSTM."
420,MaskGAN: Better Text Generation via Filling in the _______,"Neural text generation models are often autoregressive language models or seq2seq models.Neural autoregressive and seq2seq models that generate text by sampling words sequentially, with each word conditioned on the previous model, are state-of-the-art for several machine translation and summarization benchmarks.These benchmarks are often defined by validation perplexity even though this is not a direct measure of sample quality.Language models are typically trained via maximum likelihood and most often with teacher forcing.Teacher forcing is well-suited to optimizing perplexity but can result in poor sample quality because generating text requires conditioning on sequences of words that were never observed at training time.We propose to improve sample quality using Generative Adversarial Network, which explicitly train the generator to produce high quality samples and have shown a lot of success in image generation.GANs were originally to designed to output differentiable values, so discrete language generation is challenging for them.We introduce an actor-critic conditional GAN that fills in missing text conditioned on the surrounding context.We show qualitatively and quantitatively, evidence that this produces more realistic text samples compared to a maximum likelihood trained model.",Natural language GAN for filling in the blankThis paper proposes to generate text using GANs.Generating text samples using GAN and a mechanism to fill in missing words conditional on the surrounding text
421,Psychophysical vs. learnt texture representations in novelty detection,"Parametric texture models have been applied successfully to synthesize artificial images.Psychophysical studies show that under defined conditions observers are unable to differentiate between model-generated and original natural textures.In industrial applications the reverse case is of interest: a texture analysis system should decide if human observers are able to discriminate between a reference and a novel texture.For example, in case of inspecting decorative surfaces the de- tection of visible texture anomalies without any prior knowledge is required.Here, we implemented a human-vision-inspired novelty detection approach.Assuming that the features used for texture synthesis are important for human texture percep- tion, we compare psychophysical as well as learnt texture representations based on activations of a pretrained CNN in a novelty detection scenario.Additionally, we introduce a novel objective function to train one-class neural networks for novelty detection and compare the results to standard one-class SVM approaches.Our experiments clearly show the differences between human-vision-inspired texture representations and learnt features in detecting visual anomalies.Based on a dig- ital print inspection scenario we show that psychophysical texture representations are able to outperform CNN-encoded features.",Comparison of psychophysical and CNN-encoded texture representations in a one-class neural network novelty detection application.This paper focuses on novelty detection and shows that psychophysical representations can outperform VGG-encoder features in some part of this taskThis paper considers detecting anomalies in textures and proposes original loss function.Proposes training two anomaly detectors from three different models to detect perceptual anomalies in visual textures.
422,Learning Less-Overlapping Representations,"In representation learning, how to make the learned representations easy to interpret and less overfitted to training data are two important but challenging issues.To address these problems, we study a new type of regularization approach that encourages the supports of weight vectors in RL models to have small overlap, by simultaneously promoting near-orthogonality among vectors and sparsity of each vector.We apply the proposed regularizer to two models: neural networks and sparse coding, and develop an efficient ADMM-based algorithm for regularized SC.Experiments on various datasets demonstrate that weight vectors learned under our regularizer are more interpretable and have better generalization performance.","We propose a new type of regularization approach that encourages non-overlapness in representation learning, for the sake of improving interpretability and reducing overfitting.The paper introduces a matrix regularizer to simultaneously induce both sparsity and approximate orthogonality.The paper studies a regularization method to promote sparsity and reduce the overlap among the supports of the weight vectors in the learned representations to enhance interpretability and avoid overfittingThe paper proposed a new regularization approach that simultaneously encourages the weight vectors (W) to be sparse and orthogonal to each other."
423,Can recurrent neural networks warp time?,"Successful recurrent models such as long short-term memories and gated recurrent units use gating mechanisms. Empirically these models have been found to improve the learning of medium to long term temporal dependencies and to help with vanishing gradient issues.We prove that learnable gates in a recurrent model formally provide in the input data.We recover part of the LSTM architecture from a simple axiomatic approach.This result leads to a new way of initializing gate biases in LSTMs and GRUs.Experimentally, this new is shown to greatly improve learning of long term dependencies, with minimal implementation effort.","Proves that gating mechanisms provide invariance to time transformations. Introduces and tests a new initialization for LSTMs from this insight.Paper links recurrent network deisgn and its effect on how the network reacts to time transformations, and uses this to develop a simple bias initialization scheme."
424,Learning to Teach,"Teaching plays a very important role in our society, by spreading human knowledge and educating our next generations.A good teacher will select appropriate teaching materials, impact suitable methodologies, and set up targeted examinations, according to the learning behaviors of the students.In the field of artificial intelligence, however, one has not fully explored the role of teaching, and pays most attention to machine .In this paper, we argue that equal attention, if not more, should be paid to teaching, and furthermore, an optimization framework should be used to obtain good teaching strategies.We call this approach learning to teach''.In the approach, two intelligent agents interact with each other: a student model, and a teacher model.The teacher model leverages the feedback from the student model to optimize its own teaching strategies by means of reinforcement learning, so as to achieve teacher-student co-evolution.To demonstrate the practical value of our proposed approach, we take the training of deep neural networks as an example, and show that by using the learning to teach techniques, we are able to use much less training data and fewer iterations to achieve almost the same accuracy for different kinds of DNN models under various machine learning tasks.","We propose and verify the effectiveness of learning to teach, a new framework to automatically guide machine learning process.This paper focuses on ""machine teaching"" and proposes leveraging reinforcement learning by defining the reward as how fast the learner learns and using policy gradient to update the teacher parametersThe authors define a deep learning model composed of four components: a student model, a teacher model, a loss function, and a data set. Suggests a ""learning to teach"" framework, corresponding to choices over the data presented to the learner."
425,DL2: Training and Querying Neural Networks with Logic,"We present DL2, a system for training and querying neural networks with logical constraints.The key idea is to translate these constraints into a differentiable loss with desirable mathematical properties and to then either train with this loss in an iterative manner or to use the loss for querying the network for inputs subject to the constraints.We empirically demonstrate that DL2 is effective in both training and querying scenarios, across a range of constraints and data sets.","A differentiable loss for logic constraints for training and querying neural networks.A framework for turning queries over parameters and input, ouput pairs to neural networks into differentiable loss functions and an associated declarative language for specifying these queriesThis paper tackles the problem of combining logical approaches with neural networks by translating a logical formula into a non-negative loss function for a neural network."
426,Policy Optimization by Genetic Distillation ,"Genetic algorithms have been widely used in many practical optimization problems.Inspired by natural selection, operators, including mutation, crossoverand selection, provide effective heuristics for search and black-box optimization.However, they have not been shown useful for deep reinforcement learning, possiblydue to the catastrophic consequence of parameter crossovers of neural networks.Here, we present Genetic Policy Optimization, a new genetic algorithmfor sample-efficient deep policy optimization.GPO uses imitation learningfor policy crossover in the state space and applies policy gradient methods for mutation.Our experiments on MuJoCo tasks show that GPO as a genetic algorithmis able to provide superior performance over the state-of-the-art policy gradientmethods and achieves comparable or higher sample efficiency.","Genetic algorithms based approach for optimizing deep neural network policiesThe authors present an algorithm for training ensembles of policy networks that regularly mixes different policies in the ensemble together.This paper proposes a genetic algorithm inspired policy optimization method, which mimics the mutation and the crossover operators over policy networks."
427,ProxQuant: Quantized Neural Networks via Proximal Operators,"To make deep neural networks feasible in resource-constrained environments, it is beneficial to quantize models by using low-precision weights.One common technique for quantizing neural networks is the straight-through gradient method, which enables back-propagation through the quantization mapping.Despite its empirical success, little is understood about why the straight-through gradient method works.Building upon a novel observation that the straight-through gradient method is in fact identical to the well-known Nesterov’s dual-averaging algorithm on a quantization constrained optimization problem, we propose a more principled alternative approach, called ProxQuant , that formulates quantized network training as a regularized learning problem instead and optimizes it via the prox-gradient method.ProxQuant does back-propagation on the underlying full-precision vector and applies an efficient prox-operator in between stochastic gradient steps to encourage quantizedness.For quantizing ResNets and LSTMs, ProxQuant outperforms state-of-the-art results on binary quantization and is on par with state-of-the-art on multi-bit quantization.We further perform theoretical analyses showing that ProxQuant converges to stationary points under mild smoothness assumptions, whereas variants such as lazy prox-gradient method can fail to converge in the same setting.","A principled framework for model quantization using the proximal gradient method, with empirical evaluation and theoretical convergence analyses.Proposes ProxQuant method to train neural networks with quantized weights.Proposes solving binary nets and its variants using proximal gradient descent."
428,Exponentially vanishing sub-optimal local minima in multilayer neural networks,"Background: Statistical mechanics results; Choromanska et al.) suggest that local minima with high error are exponentially rare in high dimensions.However, to prove low error guarantees for Multilayer Neural Networks, previous works so far required either a heavily modified MNN model or training method, strong assumptions on the labels, or an unrealistically wide hidden layer with Omega units.Results: We examine a MNN with one hidden layer of piecewise linear units, a single output, and a quadratic loss.We prove that, with high probability in the limit of Nrightarrowinfty datapoints, the volume of differentiable regions of the empiric loss containing sub-optimal differentiable local minima is exponentially vanishing in comparison with the same volume of global minima, given standard normal input of dimension d_0=\ilde, and a more realistic number of d_1=\ilde hidden units.We demonstrate our results numerically: for example, 0% binary classification training error on CIFAR with only N/d_0 = 16 hidden neurons.","Bad"" local minima are vanishing in a multilayer neural net: a proof with more reasonable assumptions than beforeIn networks with a single hidden layer, the volume of suboptimal local minima exponentially decreases in comparison to global minima. This paper aims to answer why standard SGD based algorithms on neural network converge to 'good' solutions."
429,Evading Defenses to Transferable Adversarial Examples by Mitigating Attention Shift,"Deep neural networks are vulnerable to adversarial examples, which can mislead classifiers by adding imperceptible perturbations.An intriguing property of adversarial examples is their good transferability, making black-box attacks feasible in real-world applications.Due to the threat of adversarial attacks, many methods have been proposed to improve the robustness, and several state-of-the-art defenses are shown to be robust against transferable adversarial examples.In this paper, we identify the attention shift phenomenon, which may hinder the transferability of adversarial examples to the defense models.It indicates that the defenses rely on different discriminative regions to make predictions compared with normally trained models.Therefore, we propose an attention-invariant attack method to generate more transferable adversarial examples.Extensive experiments on the ImageNet dataset validate the effectiveness of the proposed method.Our best attack fools eight state-of-the-art defenses at an 82% success rate on average based only on the transferability, demonstrating the insecurity of the defense techniques.","We propose an attention-invariant attack method to generate more transferable adversarial examples for black-box attacks, which can fool state-of-the-art defenses with a high success rate.The paper proposes a new way of overcoming state of the art defences against adversarial attacks on CNN.This paper suggests that ""attention shift"" is a key property behind failure of adversarial attacks to transfer and propose an attention-invariant attack method"
430,"MACH: Embarrassingly parallel $K$-class classification in $O(d\log{K})$ memory and $O(K\log{K} + d\log{K})$ time, instead of $O(Kd)$","We present Merged-Averaged Classifiers via Hashing for-classification with large. Compared to traditional one-vs-all classifiers that require memory and inference cost, MACH only need memory while only requiring operation for inference.MACH is the first generic-classification algorithm, with provably theoretical guarantees, which requires memory without any assumption on the relationship between classes.MACH uses universal hashing to reduce classification with a large number of classes to few independent classification task with very small number of classes.We provide theoretical quantification of accuracy-memory tradeoff by showing the first connection between extreme classification and heavy hitters.With MACH we can train ODP dataset with 100,000 classes and 400,000 features on a single Titan X GPU, with the classification accuracy of 19.28%, which is the best-reported accuracy on this dataset.Before this work, the best performing baseline is a one-vs-all classifier that requires 40 billion parameters and achieves 9% accuracy. In contrast, MACH can achieve 9% accuracy with 480x reduction in the model size.With MACH, we also demonstrate complete training of fine-grained imagenet dataset, with 21,000 classes, on a single GPU.","How to Training 100,000 classes on a single GPUProposes an efficient hashing method MACH for softmax approximation in the context of large output space, which saves both memory and computation.A method for classification scheme for problems involving a large number of classes in a multi-class setting demonstrated on ODP and Imagenet-21K datasetsThe paper presents a hashing based scheme for reducing memory and computation time for K-way classification when K is large"
431,Backpropagation through the Void: Optimizing control variates for black-box gradient estimation,"Gradient-based optimization is the foundation of deep learning and reinforcement learning.Even when the mechanism being optimized is unknown or not differentiable, optimization using high-variance or biased gradient estimates is still often the best strategy.We introduce a general framework for learning low-variance, unbiased gradient estimators for black-box functions of random variables, based on gradients of a learned function.These estimators can be jointly trained with model parameters or policies, and are applicable in both discrete and continuous settings.We give unbiased, adaptive analogs of state-of-the-art reinforcement learning methods such as advantage actor-critic.We also demonstrate this framework for training discrete latent-variable models.",We present a general method for unbiased estimation of gradients of black-box functions of random variables. We apply this method to discrete variational inference and reinforcement learning. Suggests a new approach to performing gradient descent for blackbox optimization or training discrete latent variable models.
432,Do GANs learn the distribution? Some Theory and Empirics,"Do GANS actually learn the target distribution?The foundational paper of Goodfellow et al. suggested they do, if they were given sufficiently large deep nets, sample size, and computation time.A recent theoretical analysis in Arora et al. raised doubts whether the same holds when discriminator has bounded size.It showed that the training objective can approach its optimum value even if the generated distribution has very low support.In other words, the training objective is unable to prevent mode collapse.The current paper makes two contributions. It proposes a novel test for estimating support size using the birthday paradox of discrete probability.Using this evidence is presented that well-known GANs approaches do learn distributions of fairly low support. It theoretically studies encoder-decoder GANs architectures, which were proposed to learn more meaningful features via GANs, and consequently to also solve the mode-collapse issue.Our result shows that such encoder-decoder training objectives also cannot guarantee learning of the full distribution because they cannot prevent serious mode collapse.More seriously, they cannot prevent learning meaningless codes for data, contrary to usual intuition.","We propose a support size estimator of GANs's learned distribution to show they indeed suffer from mode collapse, and we prove that encoder-decoder GANs do not avoid the issue as well."", 'The paper attempts to estimate the size of the support for solutions produced by typical GANs experimentally. This paper proposes a clever new test based on the birthday paradox for measuring diversity in generated sample, with experiment results interpreted to mean that mode collapse is strong in a number of state-of-the-art generative models.The paper uses birthday paradox to show that some GAN architectures generate distributions with fairly low support."
433,Generating Natural Adversarial Examples,"Due to their complex nature, it is hard to characterize the ways in which machine learning models can misbehave or be exploited when deployed.Recent work on adversarial examples, i.e. inputs with minor perturbations that result in substantially different model predictions, is helpful in evaluating the robustness of these models by exposing the adversarial scenarios where they fail.However, these malicious perturbations are often unnatural, not semantically meaningful, and not applicable to complicated domains such as language.In this paper, we propose a framework to generate natural and legible adversarial examples that lie on the data manifold, by searching in semantic space of dense and continuous data representation, utilizing the recent advances in generative adversarial networks.We present generated adversaries to demonstrate the potential of the proposed approach for black-box classifiers for a wide range of applications such as image classification, textual entailment, and machine translation.We include experiments to show that the generated adversaries are natural, legible to humans, and useful in evaluating and analyzing black-box classifiers.","We propose a framework to generate “natural” adversaries against black-box classifiers for both visual and textual domains, by doing the search for adversaries in the latent semantic space.Suggests a method for creation of semantical adversary examples.Proposes a framework to generate natural adversarial examples by searching adversaries in a latent space of dense and continuous data representation "
434,Kronecker-factored Curvature Approximations for Recurrent Neural Networks,"Kronecker-factor Approximate Curvature is a 2nd-order optimization method which has been shown to give state-of-the-art performance on large-scale neural network optimization tasks. It is based on an approximation to the Fisher information matrix that makes assumptions about the particular structure of the network and the way it is parameterized.The original K-FAC method was applicable only to fully-connected networks, although it has been recently extended by Grosse & Martens to handle convolutional networks as well.In this work we extend the method to handle RNNs by introducing a novel approximation to the FIM for RNNs.This approximation works by modelling the covariance structure between the gradient contributions at different time-steps using a chain-structured linear Gaussian graphical model, summing the various cross-covariances, and computing the inverse in closed form.We demonstrate in experiments that our method significantly outperforms general purpose state-of-the-art optimizers like SGD with momentum and Adam on several challenging RNN training tasks.","We extend the K-FAC method to RNNs by developing a new family of Fisher approximations.The authors extends the K-FAC method to RNNs and presents 3 ways of approximating F, showing optimization results on 3 datasets, which outperforms ADAM in both number of updates and computation time.Proposes to extend the Kronecker-factor Appropriate Curvature optimization method to the setting of recurrent neural networks.The authors present a second-order method that is specifically designed for RNNs"
435,The Deep Weight Prior,"Bayesian inference is known to provide a general framework for incorporating prior knowledge or specific properties into machine learning models via carefully choosing a prior distribution.In this work, we propose a new type of prior distributions for convolutional neural networks, deep weight prior, that exploit generative models to encourage a specific structure of trained convolutional filters e.g., spatial correlations of weights.We define DWP in the form of an implicit distribution and propose a method for variational inference with such type of implicit priors.In experiments, we show that DWP improves the performance of Bayesian neural networks when training data are limited, and initialization of weights with samples from DWP accelerates training of conventional convolutional neural networks.","The generative model for kernels of convolutional neural networks, that acts as a prior distribution while training on new datasets.A method for modeling convolutional neural networks using a Bayes method. Proposes the 'deep weight prior': the idea is to elicit a prior on an auxilary dataset and then use that prior over the CNN filters to jump start inference for a data set of interest."", 'This paper explores learning informative priors for convolutional neural network models with similar problem domains by using autoencoders to obtain an expressive prior on the filtered weights of the trained networks."
436,Data-driven Feature Sampling for Deep Hyperspectral Classification and Segmentation,"The high dimensionality of hyperspectral imaging forces unique challenges in scope, size and processing requirements. Motivated by the potential for an in-the-field cell sorting detector, we examine a Synechocystis sp.PCC 6803 dataset wherein cells are grown alternatively in nitrogen rich or deplete cultures. We use deep learning techniques to both successfully classify cells and generate a mask segmenting the cells/condition from the background.Further, we use the classification accuracy to guide a data-driven, iterative feature selection method, allowing the design neural networks requiring 90% fewer input features with little accuracy degradation.",We applied deep learning techniques to hyperspectral image segmentation and iterative feature sampling.Proposes a greedy scheme to select a subset of highly correlated spectral features in a classification task.The paper explores the use of neural networks for classification and segmentation of hypersepctral imaging (HSI) of cells.Classifying cells and implementing cell segmentation based on deep learning techniques with reduction of input features
437,Data Interpretation and Reasoning Over Scientific Plots,"Data Interpretation is an important part of Quantitative Aptitude exams and requires an individual to answer questions grounded in plots such as bar charts, line graphs, scatter plots, .Recently, there has been an increasing interest in building models which can perform this task by learning from datasets containing triplets of the form .Two such datasets have been proposed in the recent past which contain plots generated from synthetic data with limited axes variables question templates and answer vocabulary and hence do not adequately capture the challenges posed by this task.To overcome these limitations of existing datasets, we introduce a new dataset containing million question-answer pairs grounded over plots with three main differentiators.First, the plots in our dataset contain a wide variety of realistic- variables such as CO2 emission, fertility rate, extracted from real word data sources such as World Bank, government sites, .Second, the questions in our dataset are more complex as they are based on templates extracted from interesting questions asked by a crowd of workers using a fraction of these plots.Lastly, the answers in our dataset are not restricted to a small vocabulary and a large fraction of the answers seen at test time are not present in the training vocabulary.As a result, existing models for Visual Question Answering which largely use end-to-end models in a multi-class classification framework cannot be used for this task.We establish initial results on this dataset and emphasize the complexity of the task using a multi-staged modular pipeline with various sub-components to extract relevant data from the plot and convert it to a semi-structured table combine the question with this table and use compositional semantic parsing to arrive at a logical form from which the answer can be derived.We believe that such a modular framework is the best way to go forward as it would enable the research community to independently make progress on all the sub-tasks involved in plot question answering.",We created a new dataset for data interpretation over plots and also propose a baseline for the same.The authors propose a pipeline to solve the DIP problem involving learning from datasets containing triplets of the formProposes an algorithm that can interpret data shown in scientific plots.
438,Initialization matters: Orthogonal Predictive State Recurrent Neural Networks,"Learning to predict complex time-series data is a fundamental challenge in a range of disciplines including Machine Learning, Robotics, and Natural Language Processing.Predictive State Recurrent Neural Networks are a state-of-the-art approach for modeling time-series data which combine the benefits of probabilistic filters and Recurrent Neural Networks into a single model.PSRNNs leverage the concept of Hilbert Space Embeddings of distributions to embed predictive states into a Reproducing Kernel Hilbert Space, then estimate, predict, and update these embedded states using Kernel Bayes Rule.Practical implementations of PSRNNs are made possible by the machinery of Random Features, where input features are mapped into a new space where dot products approximate the kernel well.Unfortunately PSRNNs often require a large number of RFs to obtain good results, resulting in large models which are slow to execute and slow to train.Orthogonal Random Features is an improvement on RFs which has been shown to decrease the number of RFs required for pointwise kernel approximation.Unfortunately, it is not clear that ORFs can be applied to PSRNNs, as PSRNNs rely on Kernel Ridge Regression as a core component of their learning algorithm, and the theoretical guarantees of ORF do not apply in this setting.In this paper, we extend the theory of ORFs to Kernel Ridge Regression and show that ORFs can be used to obtain Orthogonal PSRNNs, which are smaller and faster than PSRNNs.In particular, we show that OPSRNN models clearly outperform LSTMs and furthermore, can achieve accuracy similar to PSRNNs with an order of magnitude smaller number of features needed.",Improving Predictive State Recurrent Neural Networks via Orthogonal Random FeaturesProposes improving the performances of Predicitve State Recurrent Neural Networks by considering Orthogonal Random Features.The paper tackles the problem of training predictive state recurrent neural networks and makes two contributions.
439,Fidelity-Weighted Learning,"Training deep neural networks requires many training samples, but in practice training labels are expensive to obtain and may be of varying quality, as some may be from trusted expert labelers while others might be from heuristics or other sources of weak supervision such as crowd-sourcing.This creates a fundamental quality- versus-quantity trade-off in the learning process.Do we learn from the small amount of high-quality data or the potentially large amount of weakly-labeled data?We argue that if the learner could somehow know and take the label-quality into account when learning the data representation, we could get the best of both worlds.To this end, we propose “fidelity-weighted learning”, a semi-supervised student- teacher approach for training deep neural networks using weakly-labeled data.FWL modulates the parameter updates to a student network on a per-sample basis according to the posterior confidence of its label-quality estimated by a teacher.Both student and teacher are learned from the data.We evaluate FWL on two tasks in information retrieval and natural language processing where we outperform state-of-the-art alternative semi-supervised methods, indicating that our approach makes better use of strong and weak labels, and leads to better task-dependent data representations.","We propose Fidelity-weighted Learning, a semi-supervised teacher-student approach for training neural networks using weakly-labeled data.This paper suggests an approach for learning with weak supervision by using a clean and a noisy dataset and assuming a teacher and student networksThe paper attemps to train deep neural network models with few labelled training samples.The authors propose an approach for training deep learning models for situations where there is not enough reliable annotated data."
440,Online Learning for Supervised Dimension Reduction," Online learning has attracted great attention due to the increasing demand for systems that have the ability of learning and evolving.When the data to be processed is also high dimensional and dimension reduction is necessary for visualization or prediction enhancement, online dimension reduction will play an essential role.The purpose of this paper is to propose new online learning approaches for supervised dimension reduction.Our first algorithm is motivated by adapting the sliced inverse regression, a pioneer and effective algorithm for supervised dimension reduction, and making it implementable in an incremental manner.The new algorithm, called incremental sliced inverse regression, is able to update the subspace of significant factors with intrinsic lower dimensionality fast and efficiently when new observations come in.We also refine the algorithm by using an overlapping technique and develop an incremental overlapping sliced inverse regression algorithm.We verify the effectiveness and efficiency of both algorithms by simulations and real data applications.","We proposed two new approaches, the incremental sliced inverse regression and incremental overlapping sliced inverse regression, to implement supervised dimension reduction in an online learning manner.Studies sufficient dimension reduction problem and proposes an incremental sliced inverse regression algorithm.This paper proposes an online learning algorithm for supervised dimension reduction, called incremental sliced inverse regression"
441,Recursive Binary Neural Network Learning Model  with 2-bit/weight Storage Requirement,"This paper presents a storage-efficient learning model titled Recursive Binary Neural Networks for embedded and mobile devices having a limited amount of on-chip data storage such as hundreds of kilo-Bytes.The main idea of the proposed model is to recursively recycle data storage of weights during training.This enables a device with a given storage constraint to train and instantiate a neural network classifier with a larger number of weights on a chip, achieving better classification accuracy.Such efficient use of on-chip storage reduces off-chip storage accesses, improving energy-efficiency and speed of training.We verified the proposed training model with deep and convolutional neural network classifiers on the MNIST and voice activity detection benchmarks.For the deep neural network, our model achieves data storage requirement of as low as 2 bits/weight, whereas the conventional binary neural network learning models require data storage of 8 to 32 bits/weight.With the same amount of data storage, our model can train a bigger network having more weights, achieving 1% less test error than the conventional binary neural network learning model.To achieve the similar classification error, the conventional binary neural network model requires 4× more data storage for weights than our proposed model.For the convolution neural network classifier, the proposed model achieves 2.4% less test error for the same on-chip storage or 6× storage savings to achieve the similar accuracy.","We propose a learning model enabling DNN to learn with only 2 bit/weight, which is especially useful for on-device learningProposes a method to discretisize a NN incrementally to improve memory and performance."
442,Transfer Learning on Manifolds via Learned Transport Operators,"Within-class variation in a high-dimensional dataset can be modeled as being on a low-dimensional manifold due to the constraints of the physical processes producing that variation.We desire a method for learning a representation of the manifolds induced by identity-preserving transformations that can be used to increase robustness, reduce the training burden, and encourage interpretability in machine learning tasks.In particular, what is needed is a representation of the transformation manifold that can robustly capture the shape of the manifold from the input data, generate new points on the manifold, and extend transformations outside of the training domain without significantly increasing the error.Previous work has proposed algorithms to efficiently learn analytic operators that define the process of transporting one data point on a manifold to another. The main contribution of this paper is to define two transfer learning methods that use this generative manifold representation to learn natural transformations and incorporate them into new data.The first method uses this representation in a novel randomized approach to transfer learning that employs the learned generative model to map out unseen regions of the data space.These results are shown through demonstrations of transfer learning in a data augmentation task for few-shot image classification.The second method use of transport operators for injecting specific transformations into new data examples which allows for realistic image animation and informed data augmentation. These results are shown on stylized constructions using the classic swiss roll data structure and in demonstrations of transfer learning in a data augmentation task for few-shot image classification.We also propose the use of transport operators for injecting transformations into new data examples which allows for realistic image animation.","Learning transport operators on manifolds forms a valuable representation for doing tasks like transfer learning.Uses a dictionary learning framework to learn manifold transport operators on augmented USPS digits.The paper considers the framework of manifold transport operator learning of Culpepper and Olshausen (2009), and interprets it as obtaining a MAP estimate under a probabilistic generative model."
443,SCAN: Learning Hierarchical Compositional Visual Concepts,"The seemingly infinite diversity of the natural world arises from a relatively small set of coherent rules, such as the laws of physics or chemistry.We conjecture that these rules give rise to regularities that can be discovered through primarily unsupervised experiences and represented as abstract concepts.If such representations are compositional and hierarchical, they can be recombined into an exponentially large set of new concepts.This paper describes SCAN, a new framework for learning such abstractions in the visual domain.SCAN learns concepts through fast symbol association, grounding them in disentangled visual primitives that are discovered in an unsupervised manner.Unlike state of the art multimodal generative model baselines, our approach requires very few pairings between symbols and images and makes no assumptions about the form of symbol representations.Once trained, SCAN is capable of multimodal bi-directional inference, generating a diverse set of image samples from symbolic descriptions and vice versa.It also allows for traversal and manipulation of the implicit hierarchy of visual concepts through symbolic instructions and learnt logical recombination operations.Such manipulations enable SCAN to break away from its training data distribution and imagine novel visual concepts through symbolically instructed recombination of previously learnt concepts.","We present a neural variational model for learning language-guided compositional visual concepts.Proposes a novel neural net architecture that learns object concepts by combining a beta-VAE and SCAN.This paper introduces a VAE-based model for translating between images and text, with their latent representation well-suited to applying symbolic operations, giving them a more expressive language for sampling images from text. This paper proposes a new model called SCAN (Symbol-Concept Association Network) for hierarchical concept learning and allows for generalization to new concepts composed from existing concepts using logical operators."
444,Latent Topic Conversational Models,"Despite much success in many large-scale language tasks, sequence-to-sequence models have not been an ideal choice for conversational modeling as they tend to generate generic and repetitive responses.In this paper, we propose a Latent Topic Conversational Model that augments the seq2seq model with a neural topic component to better model human-human conversations.The neural topic component encodes information from the source sentence to build a global “topic” distribution over words, which is then consulted by the seq2seq model to improve generation at each time step.The experimental results show that the proposed LTCM can generate more diverse and interesting responses by sampling from its learnt latent representations.In a subjective human evaluation, the judges also confirm that LTCM is the preferred option comparing to competitive baseline models.","Latent Topic Conversational Model, a hybrid of seq2seq and neural topic model to generate more diverse and interesting responses.This paper proposed the combination of topic model and seq2seq conversational modelProposes a conversational model with topical information by combining seq2seq model with neural topic models and shows the proposed model outperforms some the baseline model seq2seq and other latent variable model variant of seq2seq.The paper addresses the issue of enduring topicality in conversation models and proposes a model which is a combination of a neural topic model and a seq2seq-based dialog system. "
445,DEEP GEOMETRICAL GRAPH CLASSIFICATION,"Most of the existing Graph Neural Networks are the mere extension of the Convolutional Neural Networks to graphs.Generally, they consist of several steps of message passing between the nodes followed by a global indiscriminate feature pooling function.In many data-sets, however, the nodes are unlabeled or their labels provide no information about the similarity between the nodes and the locations of the nodes in the graph.Accordingly, message passing may not propagate helpful information throughout the graph.We show that this conventional approach can fail to learn to perform even simple graph classification tasks.We alleviate this serious shortcoming of the GNNs by making them a two step method.In the first of the proposed approach, a graph embedding algorithm is utilized to obtain a continuous feature vector for each node of the graph.The embedding algorithm represents the graph as a point-cloud in the embedding space.In the second step, the GNN is applied to the point-cloud representation of the graph provided by the embedding method.The GNN learns to perform the given task by inferring the topological structure of the graph encoded in the spatial distribution of the embedded vectors.In addition, we extend the proposed approach to the graph clustering problem and a new architecture for graph clustering is proposed.Moreover, the spatial representation of the graph is utilized to design a graph pooling algorithm.We turn the problem of graph down-sampling into a column sampling problem, i.e., the sampling algorithm selects a subset of the nodes whose feature vectors preserve the spatial distribution of all the feature vectors.We apply the proposed approach to several popular benchmark data-sets and it is shown that the proposed geometrical approach strongly improves the state-of-the-art result for several data-sets.For instance, for the PTC data-set, we improve the state-of-the-art result for more than 22 %.",The graph analysis problem is transformed into a point cloud analysis problem. Proposes a deep GNN network for graph classification problems using their adaptive graph pooling layer.The authors propose a method for learning representations for graphs
446,Generating Adversarial Examples with Adversarial Networks,"Deep neural networks have been found to be vulnerable to adversarial examples resulting from adding small-magnitude perturbations to inputs.Such adversarial examples can mislead DNNs to produce adversary-selected results.Different attack strategies have been proposed to generate adversarial examples, but how to produce them with high perceptual quality and more efficiently requires more research efforts.In this paper, we propose AdvGAN to generate adversarial examples with generative adversarial networks, which can learn and approximate the distribution of original instances.For AdvGAN, once the generator is trained, it can generate adversarial perturbations efficiently for any instance, so as to potentially accelerate adversarial training as defenses. We apply AdvGAN in both semi-whitebox and black-box attack settings.In semi-whitebox attacks, there is no need to access the original target model after the generator is trained, in contrast to traditional white-box attacks.In black-box attacks, we dynamically train a distilled model for the black-box model and optimize the generator accordingly.Adversarial examples generated by AdvGAN on different target models have high attack success rate under state-of-the-art defenses compared to other attacks.Our attack has placed the first with 92.76% accuracy on a public MNIST black-box attack challenge.","We propose to generate adversarial example based on generative adversarial networks in a semi-whitebox and black-box settings.Describes AdvGAN, a conditional GAN plus adversarial loss, and evaluates AdvGAN on semi-white box and black box setting, reporting state-of-art results. This paper proposes a way of generating adversarial examples that fool classification systems and wins MadryLab's mnist challenge."
447,Training Deep AutoEncoders for Recommender Systems,"This paper proposes a new model for the rating prediction task in recommender systems which significantly outperforms previous state-of-the art models on a time-split Netflix data set.Our model is based on deep autoencoder with 6 layers and is trained end-to-end without any layer-wise pre-training.We empirically demonstrate that:a) deep autoencoder models generalize much better than the shallow ones,b) non-linear activation functions with negative parts are crucial for training deep models, andc) heavy use of regularization techniques such as dropout is necessary to prevent over-fitting.We also propose a new training algorithm based on iterative output re-feeding to overcome natural sparseness of collaborate filtering.The new algorithm significantly speeds up training and improves model performance.Our code is publicly available.",This paper demonstrates how to train deep autoencoders end-to-end to achieve SoA results on time-split Netflix data set.This paper presents a deep autoencoder model for rating prediction that outperforms other state-of-the-art approahces on the Netflix prize dataset. Proposes to use a deep AE to do rating prediction tasks in recommender systems.The authors present a model for more accurate Netflix recommendations demonstrating that a deep autoencoder can out-perform more complex RNN-based models that have temporal information. 
448,Universal Transformers,"Recurrent neural networks sequentially process data by updating their state with each new data point, and have long been the de facto choice for sequence modeling tasks.However, their inherently sequential computation makes them slow to train.Feed-forward and convolutional architectures have recently been shown to achieve superior results on some sequence modeling tasks such as machine translation, with the added advantage that they concurrently process all inputs in the sequence, leading to easy parallelization and faster training times.Despite these successes, however, popular feed-forward sequence models like the Transformer fail to generalize in many simple tasks that recurrent models handle with ease, e.g. copying strings or even simple logical inference when the string or formula lengths exceed those observed at training time.We propose the Universal Transformer, a parallel-in-time self-attentive recurrent sequence model which can be cast as a generalization of the Transformer model and which addresses these issues.UTs combine the parallelizability and global receptive field of feed-forward sequence models like the Transformer with the recurrent inductive bias of RNNs.We also add a dynamic per-position halting mechanism and find that it improves accuracy on several tasks.In contrast to the standard Transformer, under certain assumptions UTs can be shown to be Turing-complete.Our experiments show that UTs outperform standard Transformers on a wide range of algorithmic and language understanding tasks, including the challenging LAMBADA language modeling task where UTs achieve a new state of the art, and machine translation where UTs achieve a 0.9 BLEU improvement over Transformers on the WMT14 En-De dataset.","We introduce the Universal Transformer, a self-attentive parallel-in-time recurrent sequence model that outperforms Transformers and LSTMs on a wide range of sequence-to-sequence tasks, including machine translation.Proposes a new model UT, based on the Transformer model, with added recurrence and dynamic halting of the recurrence.This paper extends Transformer by recursively applying a multi-head self-attention block, rather than stacking multiple blocks in the vanilla Transformer."
449,Interpretable Continual Learning,"We present a framework for interpretable continual learning.We show that explanations of previously performed tasks can be used to improve performance on future tasks.ICL generates a good explanation of a finished task, then uses this to focus attention on what is important when facing a new task.The ICL idea is general and may be applied to many continual learning approaches.Here we focus on the variational continual learning framework to take advantage of its flexibility and efficacy in overcoming catastrophic forgetting.We use saliency maps to provide explanations of performed tasks and propose a new metric to assess their quality.Experiments show that ICL achieves state-of-the-art results in terms of overall continual learning performance as measured by average classification accuracy, and also in terms of its explanations, which are assessed qualitatively and quantitatively using the proposed metric.","The paper develops an interpretable continual learning framework where explanations of the finished tasks are used to enhance the attention of the learner during the future tasks, and where an explanation metric is proposed too. The authors propose a framework for continual learning based on explanations for performed classifications of previously learned tasksThis paper proposes an extension to the continual learning framework using existing variational continual learning as the base method with weight of evidence."
450,Mixed Precision Training of Convolutional Neural Networks using Integer Operations,"The state-of-the-art for mixed precision training is dominated by variants of low precision floating point operations, and in particular, FP16 accumulating into FP32 Micikevicius et al..On the other hand, while a lot of research has also happened in the domain of low and mixed-precision Integer training, these works either present results for non-SOTA networks, or relatively small datasets.In this work, we train state-of-the-art visual understanding neural networks on the ImageNet-1K dataset, with Integer operations on General Purpose hardware.In particular, we focus on Integer Fused-Multiply-and-Accumulate operations which take two pairs of INT16 operands and accumulate results into an INT32 output.We propose a shared exponent representation of tensors and develop a Dynamic Fixed Point scheme suitable for common neural network operations.The nuances of developing an efficient integer convolution kernel is examined, including methods to handle overflow of the INT32 accumulator.We implement CNN training for ResNet-50, GoogLeNet-v1, VGG-16 and AlexNet; and these networks achieve or exceed SOTA accuracy within the same number of iterations as their FP32 counterparts without any change in hyper-parameters and with a 1.8X improvement in end-to-end training throughput.To the best of our knowledge these results represent the first INT16 training results on GP hardware for ImageNet-1K dataset using SOTA CNNs and achieve highest reported accuracy using half precision","Mixed precision training pipeline using 16-bit integers on general purpose HW; SOTA accuracy for ImageNet-class CNNs; Best reported accuracy for ImageNet-1K classification task with any reduced precision training;This paper shows that a careful implementation of mixed-precision dynamic fixed point computation can achieve state of the art accuracy using a reduced precision deep learning model with a 16 bit integer representationProposes a ""dynamic fixed point"" scheme that shares the exponent part for a tensor and develops procedures to do NN computing with this format and demonstrates this for limited precision training."
451,Gated ConvNets for Letter-Based ASR,"In this paper we introduce a new speech recognition system, leveraging a simple letter-based ConvNet acoustic model.The acoustic model requires only audio transcription for training -- no alignment annotations, nor any forced alignment step is needed. At inference, our decoder takes only a word list and a language model, and is fed with letter scores from the acoustic model -- no phonetic word lexicon is needed.Key ingredients for the acoustic model are Gated Linear Units and high dropout.We show near state-of-the-art results in word error rate on the LibriSpeech corpus with MFSC features, both on the clean and other configurations.","A letter-based ConvNet acoustic model leads to a simple and competitive speech recognition pipeline.This paper applies gated convolutional neural networks to speech recognition, using the training criterion ASG."
452,IVE-GAN: Invariant Encoding Generative Adversarial Networks,"Generative adversarial networks are a powerful framework for generative tasks.However, they are difficult to train and tend to miss modes of the true data generation process.Although GANs can learn a rich representation of the covered modes of the data in their latent space, the framework misses an inverse mapping from data to this latent space.We propose Invariant Encoding Generative Adversarial Networks, a novel GAN framework that introduces such a mapping for individual samples from the data by utilizing features in the data which are invariant to certain transformations.Since the model maps individual samples to the latent space, it naturally encourages the generator to cover all modes.We demonstrate the effectiveness of our approach in terms of generative performance and learning rich representations on several datasets including common benchmark image generation tasks.","A noval GAN framework that utilizes transformation-invariant features to learn rich representations and strong generators.Proposes a modified GAN objective consisting of a classic GAN term and an invariant encoding term.This paper presents the IVE-GAN, a model that introduces en encoder to the Generative Adversarial Network framework."
453,Variational Autoencoders with Jointly Optimized Latent Dependency Structure,"We propose a method for learning the dependency structure between latent variables in deep latent variable models. Our general modeling and inference framework combines the complementary strengths of deep generative models and probabilistic graphical models.In particular, we express the latent variable space of a variational autoencoder in terms of a Bayesian network with a learned, flexible dependency structure. The network parameters, variational parameters as well as the latent topology are optimized simultaneously with a single objective. Inference is formulated via a sampling procedure that produces expectations over latent variable structures and incorporates top-down and bottom-up reasoning over latent variable values. We validate our framework in extensive experiments on MNIST, Omniglot, and CIFAR-10.Comparisons to state-of-the-art structured variational autoencoder baselines show improvements in terms of the expressiveness of the learned model.","We propose a method for learning latent dependency structure in variational autoencoders.Uses a matrix of binary random variables to capture dependencies between latent variables in a hierarchical deep generative model.This paper presents a VAE approach in which a dependency structure on the latent variable is learned during training.The authors propose to augment the latent space of a VAE with an auto-regressive structure, to improve the expressiveness of both the inference network and the latent prior"
454,Pyramid Recurrent Neural Networks for Multi-Scale Change-Point Detection,"Many real-world time series, such as in activity recognition, finance, or climate science, have changepoints where the system's structure or parameters change.Detecting changes is important as they may indicate critical events.However, existing methods for changepoint detection face challenges when the patterns of change cannot be modeled using simple and predefined metrics, and changes can occur gradually, at multiple time-scales.To address this, we show how changepoint detection can be treated as a supervised learning problem, and propose a new deep neural network architecture that can efficiently identify both abrupt and gradual changes at multiple scales.Our proposed method, pyramid recurrent neural network, is designed to be scale-invariant, by incorporating wavelets and pyramid analysis techniques from multi-scale signal processing.Through experiments on synthetic and real-world datasets, we show that PRNN can detect abrupt and gradual changes with higher accuracy than the state of the art and can extrapolate to detect changepoints at novel timescales that have not been seen in training.",We introduce a scale-invariant neural network architecture for changepoint detection in multivariate time series.The paper leverages the concept of wavelet transform within a deep architecture to solve change point detection.This paper proposes a pyramid based neural net and applies it to 1D signals with underlying processes occurring at different time scales where the task is change point detection
455,Learning Heuristics for Automated Reasoning through Reinforcement Learning,"We demonstrate how to learn efficient heuristics for automated reasoning algorithms through deep reinforcement learning.We focus on backtracking search algorithms for quantified Boolean logics, which already can solve formulas of impressive size - up to 100s of thousands of variables.The main challenge is to find a representation of these formulas that lends itself to making predictions in a scalable way.For challenging problems, the heuristic learned through our approach reduces execution time by a factor of 10 compared to the existing handwritten heuristics.","RL finds better heuristics for automated reasoning algorithms.Aims to learn a heuristic for a backtracking search algorithm utilizing Reinforcement learning and proposes a model that makes use of Graphical Neural Networks to produce literal and clauses embedding, and use them to predict the quality of each literal to decide the probability of each action.The paper proposes an approach to automatically learning variable selection heuristics for QBF using deep learning"
456,Towards a Testable Notion of Generalization for Generative Adversarial Networks,"We consider the question of how to assess generative adversarial networks, in particular with respect to whether or not they generalise beyond memorising the training data.We propose a simple procedure for assessing generative adversarial network performance based on a principled consideration of what the actual goal of generalisation is.Our approach involves using a test set to estimate the Wasserstein distance between the generative distribution produced by our procedure, and the underlying data distribution.We use this procedure to assess the performance of several modern generative adversarial network architectures.We find that this procedure is sensitive to the choice of ground metric on the underlying data space, and suggest a choice of ground metric that substantially improves performance. We finally suggest that attending to the ground metric used in Wasserstein generative adversarial network training may be fruitful, and outline a concrete pathway towards doing so.","Assess whether or not your GAN is actually doing something other than memorizing the training data.Aims to provide a quality measure/test for GANs and proposes to evaluate the current approximation of a distribution learnt by a GAN by using Wasserstein distance between two distributions made of a sum of Diracs as a baseline performance. This paper proposed a procedure for assessing the performance of GANs by re-considering the key of observation, using the procedure to test and improve the current GANs"
457,Searching for Activation Functions,"The choice of activation functions in deep networks has a significant effect on the training dynamics and task performance.Currently, the most successful and widely-used activation function is the Rectified Linear Unit.Although various hand-designed alternatives to ReLU have been proposed, none have managed to replace it due to inconsistent gains.In this work, we propose to leverage automatic search techniques to discover new activation functions.Using a combination of exhaustive and reinforcement learning-based search, we discover multiple novel activation functions.We verify the effectiveness of the searches by conducting an empirical evaluation with the best discovered activation function.Our experiments show that the best discovered activation function, f = x * sigmoid, which we name Swish, tends to work better than ReLU on deeper models across a number of challenging datasets.For example, simply replacing ReLUs with Swish units improves top-1 classification accuracy on ImageNet by 0.9% for Mobile NASNet-A and 0.6% for Inception-ResNet-v2.The simplicity of Swish and its similarity to ReLU make it easy for practitioners to replace ReLUs with Swish units in any neural network.","We use search techniques to discover novel activation functions, and our best discovered activation function, f(x) = x * sigmoid(beta * x), outperforms ReLU on a number of challenging tasks like ImageNet.Proposes a reinforcement learning based approach for finding non-linearity by searching through combinations from a set of unary and binary operators.This paper utilizes reinforcement learning to search the combination of a set of unary and binary functions resulting in a new activation functionThe author uses reinforcement learning to find new potential activation functions from a rich set of possible candidates. "
458,Building effective deep neural networks one feature at a time,"Successful training of convolutional neural networks is often associated with suffi-ciently deep architectures composed of high amounts of features.These networkstypically rely on a variety of regularization and pruning techniques to convergeto less redundant states.We introduce a novel bottom-up approach to expandrepresentations in fixed-depth architectures.These architectures start from just asingle feature per layer and greedily increase width of individual layers to attaineffective representational capacities needed for a specific task.While networkgrowth can rely on a family of metrics, we propose a computationally efficientversion based on feature time evolution and demonstrate its potency in determin-ing feature importance and a networks’ effective capacity.We demonstrate howautomatically expanded architectures converge to similar topologies that benefitfrom lesser amount of parameters or improved accuracy and exhibit systematiccorrespondence in representational complexity with the specified task.In contrastto conventional design patterns with a typical monotonic increase in the amount offeatures with increased depth, we observe that CNNs perform better when there ismore learnable parameters in intermediate, with falloffs to earlier and later layers.","A bottom-up algorithm that expands CNNs starting with one feature per layer to architectures with sufficient representational capacity.Proposes to dynamically adjust the feature map depth of a fully convolutional neural network, formulating a measure of self-resemblance and boosting performance.Introduces a simple correlation-based metric to measure whether filters in neural networks are being used effectively, as a proxy for effective capacity.Aims to address the deep learning architecture search problem via incremental addition and removal of channels in intermediate layers of the network."
459,Initialized Equilibrium Propagation for Backprop-Free Training,"Deep neural networks are almost universally trained with reverse-mode automatic differentiation.Biological networks, on the other hand, appear to lack any mechanism for sending gradients back to their input neurons, and thus cannot be learning in this way.In response to this, Scellier & Bengio proposed Equilibrium Propagation - a method for gradient-based train- ing of neural networks which uses only local learning rules and, crucially, does not rely on neurons having a mechanism for back-propagating an error gradient.Equilibrium propagation, however, has a major practical limitation: inference involves doing an iterative optimization of neural activations to find a fixed-point, and the number of steps required to closely approximate this fixed point scales poorly with the depth of the network.In response to this problem, we propose Initialized Equilibrium Propagation, which trains a feedforward network to initialize the iterative inference procedure for Equilibrium propagation.This feed-forward network learns to approximate the state of the fixed-point using a local learning rule.After training, we can simply use this initializing network for inference, resulting in a learned feedforward network.Our experiments show that this network appears to work as well or better than the original version of Equilibrium propagation.This shows how we might go about training deep networks without using backpropagation.","We train a feedforward network without backprop by using an energy-based model to provide local targetsThis paper aims at quickening the iterative inference procedure in energy-based models trained with Equilibrium Propagation (EP), by proposing to train a feedforward network to predict a fixed point of the ""equilibrating network"". Training a separate network to initialize recurrent networks trained using equilibrium propagation "
460,Adversarial Information Factorization,"We propose a novel generative model architecture designed to learn representations for images that factor out a single attribute from the rest of the representation.A single object may have many attributes which when altered do not change the identity of the object itself.Consider the human face; the identity of a particular person is independent of whether or not they happen to be wearing glasses.The attribute of wearing glasses can be changed without changing the identity of the person.However, the ability to manipulate and alter image attributes without altering the object identity is not a trivial task.Here, we are interested in learning a representation of the image that separates the identity of an object from an attribute.We demonstrate the success of our factorization approach by using the learned representation to synthesize the same face with and without a chosen attribute.We refer to this specific synthesis process as image attribute manipulation.We further demonstrate that our model achieves competitive scores, with state of the art, on a facial attribute classification task.","Learn representations for images that factor out a single attribute.This paper builds on Conditional VAE GANs to allow attribute manipulation in the synthesis process.This paper proposes a generative model to learn the representation which can separate the identity of an object from an attribute, and extends the autoencoder adversarial by adding an auxiliary network."
461,Consistent Jumpy Predictions for Videos and Scenes,"Stochastic video prediction models take in a sequence of image frames, and generate a sequence of consecutive future image frames.These models typically generate future frames in an autoregressive fashion, which is slow and requires the input and output frames to be consecutive.We introduce a model that overcomes these drawbacks by generating a latent representation from an arbitrary set of frames that can then be used to simultaneously and efficiently sample temporally consistent frames at arbitrary time-points.For example, our model can ""jump"" and directly sample frames at the end of the video, without sampling intermediate frames.Synthetic video evaluations confirm substantial gains in speed and functionality without loss in fidelity.We also apply our framework to a 3D scene reconstruction dataset.Here, our model is conditioned on camera location and can sample consistent sets of images for what an occluded region of a 3D scene might look like, even if there are multiple possibilities for what that region might contain.Reconstructions and videos are available at https://bit.ly/2O4Pc4R.","We present a model for consistent 3D reconstruction and jumpy video prediction e.g. producing image frames multiple time-steps in the future without generating intermediate frames.This paper proposes a general method for indexed data modeling by encoding index information together with observation into a neural network, and then decode the observation condition on the target index.Proposes using a VAE that encodes input video in a permuation invariant way to predict future frames of a video."
462,"Dissecting Adam: The Sign, Magnitude and Variance of Stochastic Gradients","The ADAM optimizer is exceedingly popular in the deep learning community.Often it works very well, sometimes it doesn’t.Why?We interpret ADAM as a combination of two aspects: for each weight, the update direction is determined by the sign of the stochastic gradient, whereas the update magnitude is solely determined by an estimate of its relative variance.We disentangle these two aspects and analyze them in isolation, shedding light on ADAM ’s inner workings.Transferring the ""variance adaptation” to momentum- SGD gives rise to a novel method, completing the practitioner’s toolbox for problems where ADAM fails.","Analyzing the popular Adam optimizerThe paper trys to improve Adam based on variance adaption with momentum by proposing two algorithms This paper analyzes the scale-invariance and the particular shape of the learning rate used in Adam, arguing that Adam's update is a combination of a sign-update and a variance-based learning rate."", 'The paper splits ADAM algorithm into two components: stochastic direction in sign of gradient and adaptive stepwise with relative variance, and two algorithms are proposed to test each of them."
463,Adaptive Dropout with Rademacher Complexity Regularization,"We propose a novel framework to adaptively adjust the dropout rates for the deep neural network based on a Rademacher complexity bound.The state-of-the-art deep learning algorithms impose dropout strategy to prevent feature co-adaptation.However, choosing the dropout rates remains an art of heuristics or relies on empirical grid-search over some hyperparameter space.In this work, we show the network Rademacher complexity is bounded by a function related to the dropout rate vectors and the weight coefficient matrices.Subsequently, we impose this bound as a regularizer and provide a theoretical justified way to trade-off between model complexity and representation power.Therefore, the dropout rates and the empirical loss are unified into the same objective function, which is then optimized using the block coordinate descent algorithm.We discover that the adaptively adjusted dropout rates converge to some interesting distributions that reveal meaningful patterns.Experiments on the task of image and document classification also show our method achieves better performance compared to the state-of the-art dropout algorithms.",We propose a novel framework to adaptively adjust the dropout rates for the deep neural network based on a Rademacher complexity bound.The authors connect dropout parameters to a bound of the Rademacher complexity of the network Relates complexity of networks' learnability to dropout rates in backpropagation.
464,Optimized Gated Deep Learning Architectures for Sensor Fusion,"Sensor fusion is a key technology that integrates various sensory inputs to allow for robust decision making in many applications such as autonomous driving and robot control.Deep neural networks have been adopted for sensor fusion in a body of recent studies.Among these, the so-called netgated architecture was proposed, which has demonstrated improved performances over the conventional convolu- tional neural networks.In this paper, we address several limitations of the baseline negated architecture by proposing two further optimized architectures: a coarser-grained gated architecture employing group-level fusion weights and a two-stage gated architectures leveraging both the group-level and feature- level fusion weights.Using driving mode prediction and human activity recogni- tion datasets, we demonstrate the significant performance improvements brought by the proposed gated architectures and also their robustness in the presence of sensor noise and failures.","Optimized gated deep learning architectures for sensor fusion is proposed.The authors improve upon several limitations of the baseline negated architecture by proposing a coarser-grained gated fusion architecture and a two-stage gated fusion architectureProposes two gated deep learning architectures for sensor fusion and by having the grouped features, demonstrates improved performance, especially in the presence of random sensor noise and failures."
465,A Mean Field Theory of Batch Normalization,"We develop a mean field theory for batch normalization in fully-connected feedforward neural networks.In so doing, we provide a precise characterization of signal propagation and gradient backpropagation in wide batch-normalized networks at initialization.Our theory shows that gradient signals grow exponentially in depth and that these exploding gradients cannot be eliminated by tuning the initial weight variances or by adjusting the nonlinear activation function.Indeed, batch normalization itself is the cause of gradient explosion.As a result, vanilla batch-normalized networks without skip connections are not trainable at large depths for common initialization schemes, a prediction that we verify with a variety of empirical simulations.While gradient explosion cannot be eliminated, it can be reduced by tuning the network close to the linear regime, which improves the trainability of deep batch-normalized networks without residual connections.Finally, we investigate the learning dynamics of batch-normalized networks and observe that after a single step of optimization the networks achieve a relatively stable equilibrium in which gradients have dramatically smaller dynamic range.Our theory leverages Laplace, Fourier, and Gegenbauer transforms and we derive new identities that may be of independent interest.",Batch normalization causes exploding gradients in vanilla feedforward networks.Develops a mean field theory for batch normalization (BN) in fully-connected networks with randomly initialized weights.Provides a dynamic perspective on deep neural network using the evolution of the covariance matrix along with the layers.
466,Learning a SAT Solver from Single-Bit Supervision,"We present NeuroSAT, a message passing neural network that learns to solve SAT problems after only being trained as a classifier to predict satisfiability. Although it is not competitive with state-of-the-art SAT solvers, NeuroSAT can solve problems that are substantially larger and more difficult than it ever saw during training by simply running for more iterations.Moreover, NeuroSAT generalizes to novel distributions; after training only on random SAT problems, at test time it can solve SAT problems encoding graph coloring, clique detection, dominating set, and vertex cover problems, all on a range of distributions over small random graphs.","We train a graph network to predict boolean satisfiability and show that it learns to search for solutions, and that the solutions it finds can be decoded from its activations.The paper describes a general neural network architecture for predicting satisfiabilityThis paper presents the NeuroSAT architecture which uses a deep message passing neural net for predicting the satisfiability of CNF instances"
467,Diffusion Convolutional Recurrent Neural Network: Data-Driven Traffic Forecasting,"Spatiotemporal forecasting has various applications in neuroscience, climate and transportation domain.Traffic forecasting is one canonical example of such learning task.The task is challenging due to complex spatial dependency on road networks, non-linear temporal dynamics with changing road conditions and inherent difficulty of long-term forecasting.To address these challenges, we propose to model the traffic flow as a diffusion process on a directed graph and introduce Diffusion Convolutional Recurrent Neural Network, a deep learning framework for traffic forecasting that incorporates both spatial and temporal dependency in the traffic flow.Specifically, DCRNN captures the spatial dependency using bidirectional random walks on the graph, and the temporal dependency using the encoder-decoder architecture with scheduled sampling.We evaluate the framework on two real-world large-scale road network traffic datasets and observe consistent improvement of 12% - 15% over state-of-the-art baselines",A neural sequence model that learns to forecast on a directed graph.The paper proposes the Diffusion Convolutional Recurrent Neural Network architecture for the spatiotemporal traffic forecasting problemProposes to build a traffic forecasting model using a diffusion process for convolutional recurrent neural networks to address saptio-temporal autocorrelation.
468,Reliable Uncertainty Estimates in Deep Neural Networks using Noise Contrastive Priors,"Obtaining reliable uncertainty estimates of neural network predictions is a long standing challenge.Bayesian neural networks have been proposed as a solution, but it remains open how to specify their prior.In particular, the common practice of a standard normal prior in weight space imposes only weak regularities, causing the function posterior to possibly generalize in unforeseen ways on inputs outside of the training distribution.We propose noise contrastive priors to obtain reliable uncertainty estimates.The key idea is to train the model to output high uncertainty for data points outside of the training distribution.NCPs do so using an input prior, which adds noise to the inputs of the current mini batch, and an output prior, which is a wide distribution given these inputs.NCPs are compatible with any model that can output uncertainty estimates, are easy to scale, and yield reliable uncertainty estimates throughout training.Empirically, we show that NCPs prevent overfitting outside of the training distribution and result in uncertainty estimates that are useful for active learning.We demonstrate the scalability of our method on the flight delays data set, where we significantly improve upon previously published results.",We train neural networks to be uncertain on noisy inputs to avoid overconfident predictions outside of the training distribution.Presents an approach to obtain uncertainty estimates for neural network predictions that has good performance when quantifying predictive uncertainty at points that are outside of the training distribution.The paper considers the problem of uncertainty estimation of neural networks and proposes to use Bayesian approach with noice contrastive prior
469,What a difference a pixel makes: An empirical examination of features used by CNNs for categorisation,"Convolutional neural networks were inspired by human vision and, in some settings, achieve a performance comparable to human object recognition.This has lead to the speculation that both systems use similar mechanisms to perform recognition.In this study, we conducted a series of simulations that indicate that there is a fundamental difference between human vision and CNNs: while object recognition in humans relies on analysing shape, CNNs do not have such a shape-bias.We teased apart the type of features selected by the model by modifying the CIFAR-10 dataset so that, in addition to containing objects with shape, the images concurrently contained non-shape features, such as a noise-like mask.When trained on these modified set of images, the model did not show any bias towards selecting shapes as features.Instead it relied on whichever feature allowed it to perform the best prediction -- even when this feature was a noise-like mask or a single predictive pixel amongst 50176 pixels.We also found that regularisation methods, such as batch normalisation or Dropout, did not change this behaviour and neither did past or concurrent experience with images from other datasets.","This study highlights a key difference between human vision and CNNs: while object recognition in humans relies on analysing shape, CNNs do not have such a shape-bias.Seeks to establish via a series of well-designed experiments that CNNs trained for image classification don’t encode shape-bias like human vision.This paper highlights the fact that CNNs will not necessarily learn to recognize objects based on their shape and shows they will overfeat to noise based features."
470,Multi-View Data Generation Without View Supervision,"The development of high-dimensional generative models has recently gained a great surge of interest with the introduction of variational auto-encoders and generative adversarial neural networks.Different variants have been proposed where the underlying latent space is structured, for example, based on attributes describing the data to generate.We focus on a particular problem where one aims at generating samples corresponding to a number of objects under various views.We assume that the distribution of the data is driven by two independent latent factors: the content, which represents the intrinsic features of an object, and the view, which stands for the settings of a particular observation of that object.Therefore, we propose a generative model and a conditional variant built on such a disentangled latent space.This approach allows us to generate realistic samples corresponding to various objects in a high variety of views.Unlike many multi-view approaches, our model doesn't need any supervision on the views but only on the content.Compared to other conditional generation approaches that are mostly based on binary or categorical attributes, we make no such assumption about the factors of variations.Our model can be used on problems with a huge, potentially infinite, number of categories.We experiment it on four images datasets on which we demonstrate the effectiveness of the model and its ability to generalize.","We describe a novel multi-view generative model that can generate multiple views of the same object, or multiple objects in the same view with no need of label on views.This paper presents a GAN-based method for image generation that attempts to separate latent variables describing image content from those describing properties of view.This paper proposes a GAN architecture that aims at decomposing the underlying distribution of a particular class into ""content"" and ""view"".Proposes a new generative model based on the Generative Adversarial Network (GAN) that disentangles the content and the view of objects without view supervision and extends GMV into a conditional generative model that takes an input image and generates different views of the object in the input image. "
471,An inference-based policy gradient method for learning options,"In the pursuit of increasingly intelligent learning systems, abstraction plays a vital role in enabling sophisticated decisions to be made in complex environments.The options framework provides formalism for such abstraction over sequences of decisions. However most models require that options be given a priori, presumably specified by hand, which is neither efficient, nor scalable.Indeed, it is preferable to learn options directly from interaction with the environment.Despite several efforts, this remains a difficult problem: many approaches require access to a model of the environmental dynamics, and inferred options are often not interpretable, which limits our ability to explain the system behavior for verification or debugging purposes. In this work we develop a novel policy gradient method for the automatic learning of policies with options. This algorithm uses inference methods to simultaneously improve all of the options available to an agent, and thus can be employed in an off-policy manner, without observing option labels.Experimental results show that the options learned can be interpreted.Further, we find that the method presented here is more sample efficient than existing methods, leading to faster and more stable learning of policies with options.","We develop a novel policy gradient method for the automatic learning of policies with options using a differentiable inference step.The paper presents a new policy gradient technique for learning options, where a single sample can be used to update all options.Proposes an off-policy method for learning options in complex continuous problems."
472,Still not systematic after all these years: On the compositional skills of sequence-to-sequence recurrent networks,"Humans can understand and produce new utterances effortlessly, thanks to their systematic compositional skills.Once a person learns the meaning of a new verb ""dax,"" he or she can immediately understand the meaning of ""dax twice"" or ""sing and dax.""In this paper, we introduce the SCAN domain, consisting of a set of simple compositional navigation commands paired with the corresponding action sequences.We then test the zero-shot generalization capabilities of a variety of recurrent neural networks trained on SCAN with sequence-to-sequence methods.We find that RNNs can generalize well when the differences between training and test commands are small, so that they can apply ""mix-and-match"" strategies to solve the task.However, when generalization requires systematic compositional skills, RNNs fail spectacularly.We conclude with a proof-of-concept experiment in neural machine translation, supporting the conjecture that lack of systematicity is an important factor explaining why neural networks need very large training sets.","Using a simple language-driven navigation task, we study the compositional capabilities of modern seq2seq recurrent networks.This paper focuses on the zero-shot learning compositional capabilities of modern sequence-to-sequence RNNs and exposes the short-comings of current seq2seq RNN architectures.The paper analyzes the composition abilities of RNNs, specifically, the generalization ability of RNNs on random subset of SCAN commands, on longer SCAN commands, and of composition over primitive commands. The authors introduce a new dataset that facilitates the analysis of a Seq2Seq learning case"
473,Graph Matching Networks for Learning the Similarity of Graph Structured Objects,"This paper addresses the challenging problem of retrieval and matching of graph structured objects, and makes two key contributions.First, we demonstrate how Graph Neural Networks, which have emerged as an effective model for various supervised prediction problems defined on structured data, can be trained to produce embedding of graphs in vector spaces that enables efficient similarity reasoning.Second, we propose a novel Graph Matching Network model that, given a pair of graphs as input, computes a similarity score between them by jointly reasoning on the pair through a new cross-graph attention-based matching mechanism.We demonstrate the effectiveness of our models on different domains including the challenging problem of control-flow-graph based function similarity search that plays an important role in the detection of vulnerabilities in software systems.The experimental analysis demonstrates that our models are not only able to exploit structure in the context of similarity learning but they can also outperform domain-specific baseline systems that have been carefully hand-engineered for these problems.","We tackle the problem of similarity learning for structured objects with applications in particular in computer security, and propose a new model graph matching networks that excels on this task.Authors introduce a Graph Matching Network for retrival and matching of graph structured objects.The authors attack the problem of graph matching by proposing an extension of graph embedding networksThe authors present two methods for learning a similarity score between pairs of graphs and show the beneficiality of introducing idesa from graph matching to graph neural networks."
474,Exploring Asymmetric Encoder-Decoder Structure for Context-based Sentence Representation Learning,"Context information plays an important role in human language understanding, and it is also useful for machines to learn vector representations of language.In this paper, we explore an asymmetric encoder-decoder structure for unsupervised context-based sentence representation learning.As a result, we build an encoder-decoder architecture with an RNN encoder and a CNN decoder, and we show that neither an autoregressive decoder nor an RNN decoder is required. We further combine a suite of effective designs to significantly improve model efficiency while also achieving better performance.Our model is trained on two different large unlabeled corpora, and in both cases transferability is evaluated on a set of downstream language understanding tasks.We empirically show that our model is simple and fast while producing rich sentence representations that excel in downstream tasks.","We proposed an RNN-CNN encoder-decoder model for fast unsupervised sentence representation learning.Modifications to the skip-thought framework for learning sentence embeddings.This paper presents a new RNN encoder–CNN decoder hybrid design for use in pretraining, which does not require an autoregressive decoder when pretraining encoders.The authors extend Skip-thought by decoding only one target sentence using a CNN decoder."
475,Entropic GANs meet VAEs: A Statistical Approach to Compute Sample Likelihoods in GANs,"Building on the success of deep learning, two modern approaches to learn a probability model of the observed data are Generative Adversarial Networks and Variational AutoEncoders.VAEs consider an explicit probability model for the data and compute a generative distribution by maximizing a variational lower-bound on the log-likelihood function.GANs, however, compute a generative model by minimizing a distance between observed and generated probability distributions without considering an explicit model for the observed data.The lack of having explicit probability models in GANs prohibits computation of sample likelihoods in their frameworks and limits their use in statistical inference problems.In this work, we show that an optimal transport GAN with the entropy regularization can be viewed as a generative model that maximizes a lower-bound on average sample likelihoods, an approach that VAEs are based on.In particular, our proof constructs an explicit probability model for GANs that can be used to compute likelihood statistics within GAN's framework.Our numerical results on several datasets demonstrate consistent trends with the proposed theory.",A statistical approach to compute sample likelihoods in Generative Adversarial NetworksShow that WGAN with entropic regularization maximizes a lower bound on the likelihood of the observed data distribution. Authors claim it is possible to leverage the upper bound from an entropy regularized optimal transport to come up with a measure of 'sample likelihood'.
476,Geomstats: a Python Package for Riemannian Geometry in Machine Learning,"We introduce geomstats, a Python package for Riemannian modelization and optimization over manifolds such as hyperspheres, hyperbolic spaces, SPD matrices or Lie groups of transformations.Our contribution is threefold.First, geomstats allows the flexible modeling of many a machine learning problem through an efficient and extensively unit-tested implementations of these manifolds, as well as the set of useful Riemannian metrics, exponential and logarithm maps that we provide.Moreover, the wide choice of loss functions and our implementation of the corresponding gradients allow fast and easy optimization over manifolds.Finally, geomstats is the only package to provide a unified framework for Riemannian geometry, as the operations implemented in geomstats are available with different computing backends, as well as with a GPU-enabled mode–-thus considerably facilitating the application of Riemannian geometry in machine learning.In this paper, we present geomstats through a review of the utility and advantages of manifolds in machine learning, using the concrete examples that they span to show the efficiency and practicality of their implementation using our package","We introduce geomstats, an efficient Python package for Riemannian modelization and optimization over manifolds compatible with both numpy and tensorflow .The paper introduces the software package geomstats, which provides simple use of Riemannian manifolds and metrics within machine learning modelsProposes a Python package for optimization and applications on Reimannian manifolds and highlights the differences between Geomstats package and other packages.Introduces a geometric toolbox, Geomstats, for machine learning on Reimannian manifolds."
477,Dynamic Sparse Graph for Efficient Deep Learning,"We propose to execute deep neural networks with dynamic and sparse graph structure for compressive memory and accelerative execution during both training and inference.The great success of DNNs motivates the pursuing of lightweight models for the deployment onto embedded devices.However, most of the previous studies optimize for inference while neglect training or even complicate it.Training is far more intractable, since the neurons dominate the memory cost rather than the weights in inference; the dynamic activation makes previous sparse acceleration via one-off optimization on fixed weight invalid; batch normalization is critical for maintaining accuracy while its activation reorganization damages the sparsity.To address these issues, DSG activates only a small amount of neurons with high selectivity at each iteration via a dimensionreduction search and obtains the BN compatibility via a double-mask selection.Experiments show significant memory saving and operation reduction with little accuracy loss on various benchmarks.",We construct dynamic sparse graph via dimension-reduction search to reduce compute and memory cost in both DNN training and inference.The authors propose to use dynamic sparse computation graph for reducing the computation memory and time cost in deep neural network (DNN).This paper proposes a method to speed up training and inference of deep neural networks using dynamic pruning of the compute graph.
478,Information-Directed Exploration for Deep Reinforcement Learning,"Efficient exploration remains a major challenge for reinforcement learning.One reason is that the variability of the returns often depends on the current state and action, and is therefore heteroscedastic.Classical exploration strategies such as upper confidence bound algorithms and Thompson sampling fail to appropriately account for heteroscedasticity, even in the bandit setting.Motivated by recent findings that address this issue in bandits, we propose to use Information-Directed Sampling for exploration in reinforcement learning.As our main contribution, we build on recent advances in distributional reinforcement learning and propose a novel, tractable approximation of IDS for deep Q-learning.The resulting exploration strategy explicitly accounts for both parametric uncertainty and heteroscedastic observation noise.We evaluate our method on Atari games and demonstrate a significant improvement over alternative approaches.","We develop a practical extension of Information-Directed Sampling for Reinforcement Learning, which accounts for parametric uncertainty and heteroscedasticity in the return distribution for exploration.The authors propose a way of extending Information-Directed Sampling to reinforcement learning by combining two types of uncertainty to obtain a simple exploration strategy based on IDS. This paper investigates sophistical exploration approaches for reinforcement learning built on Information Direct Sampling and on Distributional Reinforcement Learning"
479,NerveNet: Learning Structured Policy with Graph Neural Networks,"We address the problem of learning structured policies for continuous control.In traditional reinforcement learning, policies of agents are learned by MLPs which take the concatenation of all observations from the environment as input for predicting actions.In this work, we propose NerveNet to explicitly model the structure of an agent, which naturally takes the form of a graph.Specifically, serving as the agent's policy network, NerveNet first propagates information over the structure of the agent and then predict actions for different parts of the agent.In the experiments, we first show that our NerveNet is comparable to state-of-the-art methods on standard MuJoCo environments.We further propose our customized reinforcement learning environments for benchmarking two types of structure transfer learning tasks, i.e., size and disability transfer.We demonstrate that policies learned by NerveNet are significantly better than policies learned by other models and are able to transfer even in a zero-shot setting.","using graph neural network to model structural information of the agents to improve policy and transferability A method for representing and learning structured policy for continuous control tasks using Graph Neural Networks The submission proposes incorporation of additional structure into reinforcement learning problems, particularly the structure of the agent's morphology"", 'Propose an application of Graph Neural Networks to learning policies for controlling ""centipede"" robots of different lengths."
480,Hierarchical Reinforcement Learning via Advantage-Weighted Information Maximization,"Real-world tasks are often highly structured.Hierarchical reinforcement learning has attracted research interest as an approach for leveraging the hierarchical structure of a given task in reinforcement learning.However, identifying the hierarchical policy structure that enhances the performance of RL is not a trivial task.In this paper, we propose an HRL method that learns a latent variable of a hierarchical policy using mutual information maximization.Our approach can be interpreted as a way to learn a discrete and latent representation of the state-action space.To learn option policies that correspond to modes of the advantage function, we introduce advantage-weighted importance sampling. In our HRL method, the gating policy learns to select option policies based on an option-value function, and these option policies are optimized based on the deterministic policy gradient method.This framework is derived by leveraging the analogy between a monolithic policy in standard RL and a hierarchical policy in HRL by using a deterministic option policy. Experimental results indicate that our HRL approach can learn a diversity of options and that it can enhance the performance of RL in continuous control tasks.",This paper presents a hierarchical reinforcement learning framework based on deterministic option policies and mutual information maximization. Proposes an HRL algorithm that attempts to learn options that maximize their mutual information with the state-action density under the optimal policy.This paper proposes an HRL system in which the mutal information of the latent variable and the state-action pairs is approximately maximized.Proposes a criterion that aims to maximize the mutual information between options and state-action pairs and show empirically that the learned options decompose the state-action space but not the state space. 
481,Hierarchical interpretations for neural network predictions,"Deep neural networks have achieved impressive predictive performance due to their ability to learn complex, non-linear relationships between variables.However, the inability to effectively visualize these relationships has led to DNNs being characterized as black boxes and consequently limited their applications.To ameliorate this problem, we introduce the use of hierarchical interpretations to explain DNN predictions through our proposed method: agglomerative contextual decomposition.Given a prediction from a trained DNN, ACD produces a hierarchical clustering of the input features, along with the contribution of each cluster to the final prediction.This hierarchy is optimized to identify clusters of features that the DNN learned are predictive.We introduce ACD using examples from Stanford Sentiment Treebank and ImageNet, in order to diagnose incorrect predictions, identify dataset bias, and extract polarizing phrases of varying lengths.Through human experiments, we demonstrate that ACD enables users both to identify the more accurate of two DNNs and to better trust a DNN's outputs."", ""We also find that ACD's hierarchy is largely robust to adversarial perturbations, implying that it captures fundamental aspects of the input and ignores spurious noise.","We introduce and validate hierarchical local interpretations, the first technique to automatically search for and display important interactions for individual predictions made by LSTMs and CNNs.A novel approach to explain neural network predictions by learning hierarchical representations of groups of input features and their contribution to the final predictionExtends an existing feature interpretation method for LSTMs to more generic DNNs and introduces a hierarchical clustering of the input features and the contributions of each cluster to the final prediction.This paper proposes a hierarchical extension of contextual decomposition."
482,NETWORK COMPRESSION USING CORRELATION ANALYSIS OF LAYER RESPONSES,"Principal Filter Analysis is an easy to implement, yet effective method for neural network compression.PFA exploits the intrinsic correlation between filter responses within network layers to recommend a smaller network footprint.We propose two compression algorithms: the first allows a user to specify the proportion of the original spectral energy that should be preserved in each layer after compression, while the second is a heuristic that leads to a parameter-free approach that automatically selects the compression used at each layer.Both algorithms are evaluated against several architectures and datasets, and we show considerable compression rates without compromising accuracy, e.g., for VGG-16 on CIFAR-10, CIFAR-100 and ImageNet, PFA achieves a compression rate of 8x, 3x, and 1.4x with an accuracy gain of 0.4%, 1.4% points, and 2.4% respectively.In our tests we also demonstrate that networks compressed with PFA achieve an accuracy that is very close to the empirical upper bound for a given compression ratio.Finally, we show how PFA is an effective tool for simultaneous compression and domain adaptation.","We propose an easy to implement, yet effective method for neural network compression. PFA exploits the intrinsic correlation between filter responses within network layers to recommend a smaller network footprints.Proposes to prune convolutional networks by analyzing the observed correlation between the filters of a same layer as expressed by the eigenvalue spectrum of their covariance matrix.This paper introduces an approach to compressing neural networks by looking at the correlation of filter responses in each layer via two strategies.This paper proposes a compression method based on spectral analysis"
483,Learning to Coordinate Multiple Reinforcement Learning Agents for Diverse Query Reformulation,"We propose a method to efficiently learn diverse strategies in reinforcement learning for query reformulation in the tasks of document retrieval and question answering.In the proposed framework an agent consists of multiple specialized sub-agents and a meta-agent that learns to aggregate the answers from sub-agents to produce a final answer.Sub-agents are trained on disjoint partitions of the training data, while the meta-agent is trained on the full training set.Our method makes learning faster, because it is highly parallelizable, and has better generalization performance than strong baselines, such as an ensemble of agents trained on the full data.We show that the improved performance is due to the increased diversity of reformulation strategies.","Multiple diverse query reformulation agents trained with reinforcement learning to improve search engines.Parellelization of ensemble method in reinforement learning for query reformulation, speeding up training and improving the diversity of learnt freformulationsThe authors propose to train multiple distinct agents, each over a different subset of the training set.The authors propose an ensemble approach for query reformulation"
484,Conditional Network Embeddings,"Network Embeddings map the nodes of a given network into-dimensional Euclidean space.Ideally, this mapping is such that 'similar' nodes are mapped onto nearby points, such that the NE can be used for purposes such as link prediction or classification.In recent years various methods for NE have been introduced, all following a similar strategy: defining a notion of similarity between nodes, a distance measure in the embedding space, and a loss function that penalizes large distances for similar nodes and small distances for dissimilar nodes.A difficulty faced by existing methods is that certain networks are fundamentally hard to embed due to their structural properties: multipartiteness, certain degree distributions, assortativity, etc.To overcome this, we introduce a conceptual innovation to the NE literature and propose to create ; embeddings that maximally add information with respect to given structural properties.We use a simple Bayesian approach to achieve this, and propose a block stochastic gradient descent algorithm for fitting it efficiently.We demonstrate that CNEs are superior for link prediction and multi-label classification when compared to state-of-the-art methods, and this without adding significant mathematical or computational complexity.Finally, we illustrate the potential of CNE for network visualization.","We introduce a network embedding method that accounts for prior information about the network, yielding superior empirical performance.The paper proposed to use a prior distribution to constraint the network embedding, for the formulation this paper used very restricted Gaussian distributions.Proposes learning unsupervised node embeddings by considering the structural properties of networks."
485,On the Convergence of A Class of Adam-Type Algorithms  for Non-Convex Optimization,"This paper studies a class of adaptive gradient based momentum algorithms that update the search directions and learning rates simultaneously using past gradients.This class, which we refer to as the ''Adam-type'', includes the popular algorithms such as Adam, AMSGrad, AdaGrad.Despite their popularity in training deep neural networks, the convergence of these algorithms for solving non-convex problems remains an open question.In this paper, we develop an analysis framework and a set of mild sufficient conditions that guarantee the convergence of the Adam-type methods, with a convergence rate of order for non-convex stochastic optimization.Our convergence analysis applies to a new algorithm called AdaFom.We show that the conditions are essential, by identifying concrete examples in which violating the conditions makes an algorithm diverge.Besides providing one of the first comprehensive analysis for Adam-type methods in the non-convex setting, our results can also help the practitioners to easily monitor the progress of algorithms and determine their convergence behavior.","We analyze convergence of Adam-type algorithms and provide mild sufficient conditions to guarantee their convergence, we also show violating the conditions can makes an algorithm diverge.Presents a convergence analysis in the non-convex setting for a family of optimization algorithms.This paper investigates the convergence condition of Adam-type optimizers in the unconstrained non-convex optimization problems."
486,AANN: Absolute Artificial Neural Network,"This research paper describes a simplistic architecture named as AANN: Absolute Artificial Neural Network, which can be used to create highly interpretable representations of the input data.These representations are generated by penalizing the learning of the network in such a way that those learned representations correspond to the respective labels present in the labelled dataset used for supervised training; thereby, simultaneously giving the network the ability to classify the input data.The network can be used in the reverse direction to generate data that closely resembles the input by feeding in representation vectors as required.This research paper also explores the use of mathematical abs functions as activation functions which constitutes the core part of this neural network architecture.Finally the results obtained on the MNIST dataset by using this technique are presented and discussed in brief.","Tied weights auto-encoder with abs function as activation function, learns to do classification in the forward direction and regression in the backward direction due to specially defined cost function.The paper proposes using the absolute value activation function in an autoencoder architecture with an additional supervised learning term in the objective functionThis paper introduces a reversible network with absolute value used as the activation function"
487,Improving Relation Extraction by Pre-trained Language Representations,"Current state-of-the-art relation extraction methods typically rely on a set of lexical, syntactic, and semantic features, explicitly computed in a pre-processing step.Training feature extraction models requires additional annotated language resources, which severely restricts the applicability and portability of relation extraction to novel languages.Similarly, pre-processing introduces an additional source of error.To address these limitations, we introduce TRE, a Transformer for Relation Extraction, extending the OpenAI Generative Pre-trained Transformer [Radford et al., 2018].Unlike previous relation extraction models, TRE uses pre-trained deep language representations instead of explicit linguistic features to inform the relation classification and combines it with the self-attentive Transformer architecture to effectively model long-range dependencies between entity mentions.TRE allows us to learn implicit linguistic features solely from plain text corpora by unsupervised pre-training, before fine-tuning the learned language representations on the relation extraction task.TRE obtains a new state-of-the-art result on the TACRED and SemEval 2010 Task 8 datasets, achieving a test F1 of 67.4 and 87.1, respectively.Furthermore, we observe a significant increase in sample efficiency.With only 20% of the training examples, TRE matches the performance of our baselines and our model trained from scratch on 100% of the TACRED dataset.We open-source our trained models, experiments, and source code.","We propose a Transformer based relation extraction model that uses pre-trained language representations instead of explicit linguistic features.Presents a transformer-based relation extraction model that leverages pre-training on unlabeled text with a language modeling objective.This article describes a novel application of Transformer networks for relation extraction.The paper presents a Transformer based architecture for relaxation extraction, evaluating on two datasets."
488,Simple and efficient architecture search for Convolutional Neural Networks,"Neural networks have recently had a lot of success for many tasks.However, neuralnetwork architectures that perform well are still typically designed manuallyby experts in a cumbersome trial-and-error process.We propose a new methodto automatically search for well-performing CNN architectures based on a simplehill climbing procedure whose operators apply network morphisms, followedby short optimization runs by cosine annealing.Surprisingly, this simple methodyields competitive results, despite only requiring resources in the same order ofmagnitude as training a single network.E.g., on CIFAR-10, our method designsand trains networks with an error rate below 6% in only 12 hours on a single GPU;training for one day reduces this error further, to almost 5%.",We propose a simple and efficent method for architecture search for convolutional neural networks.Proposes a neural architecture search method that achieves close to state-of-the-art accuracy on CIFAR10 and takes much less computational resources.Presents a method to search neural network architectures at the same time of training which dramatically saves training time and architecture searching time.Proposes variant of neural architecture search using network morphisms to define a search space using CNN architectures completing the CIFAR image classification task
489,GraphGAN: Generating Graphs via Random Walks,"We propose GraphGAN - the first implicit generative model for graphs that enables to mimic real-world networks.We pose the problem of graph generation as learning the distribution of biased random walks over a single input graph.Our model is based on a stochastic neural network that generates discrete output samples, and is trained using the Wasserstein GAN objective.GraphGAN enables us to generate sibling graphs, which have similar properties yet are not exact replicas of the original graph.Moreover, GraphGAN learns a semantic mapping from the latent input space to the generated graph's properties.We discover that sampling from certain regions of the latent space leads to varying properties of the output graphs, with smooth transitions between them.Strong generalization properties of GraphGAN are highlighted by its competitive performance in link prediction as well as promising results on node classification, even though not specifically trained for these tasks.","Using GANs to generate graphs via random walks.The authors proposed a generative model of random walks on graphs that allows for model-agnostic learning, controllable fitting, ensemble graph generationProposes a WGAN formulation for generating graphs based on random walks using node embeddings and an LSTM architecture for modeling."
490,Novelty Detection with GAN,"The ability of a classifier to recognize unknown inputs is important for many classification-based systems.We discuss the problem of simultaneous classification and novelty detection, i.e. determining whether an input is from the known set of classes and from which specific class, or from an unknown domain and does not belong to any of the known classes.We propose a method based on the Generative Adversarial Networks framework.We show that a multi-class discriminator trained with a generator that generates samples from a mixture of nominal and novel data distributions is the optimal novelty detector.We approximate that generator with a mixture generator trained with the Feature Matching loss and empirically show that the proposed method outperforms conventional methods for novelty detection.Our findings demonstrate a simple, yet powerful new application of the GAN framework for the task of novelty detection.",We propose to solve a problem of simultaneous classification and novelty detection within the GAN framework.Proposes a GAN to unify classification and novelty detection.The paper presents a method for novelty detection based on a multi-class GAN which is trained to output images generated from a mixture of the nominal and novel distributions.The paper proposes a GAN for novelty detection using a mixture generator with feature matching loss
491,SpeakerGAN: Recognizing Speakers in New Languages with Generative Adversarial Networks," Verifying a person's identity based on their voice is a challenging, real-world problem in biometric security.A crucial requirement of such speaker verification systems is to be domain robust.Performance should not degrade even if speakers are talking in languages not seen during training.To this end, we present a flexible and interpretable framework for learning domain invariant speaker embeddings using Generative Adversarial Networks.We combine adversarial training with an angular margin loss function, which encourages the speaker embedding model to be discriminative by directly optimizing for cosine similarity between classes.We are able to beat a strong baseline system using a cosine distance classifier and a simple score-averaging strategy.Our results also show that models with adversarial adaptation perform significantly better than unadapted models.In an attempt to better understand this behavior, we quantitatively measure the degree of invariance induced by our proposed methods using Maximum Mean Discrepancy and Frechet distances.Our analysis shows that our proposed adversarial speaker embedding models significantly reduce the distance between source and target data distributions, while performing similarly on the former and better on the latter.","Speaker verificaiton performance can be significantly improved by adapting the model to in-domain data using Generative Adversarial Networks. Furthermore, the adaptation can be performed in an unsupervised way.Propose a number of GAN variants on the task of speaker recognition in the domain mismatched condition."
492,Non-Synergistic Variational Autoencoders,"Learning disentangling representations of the independent factors of variations that explain the data in an unsupervised setting is still a major challenge.In the following paper we address the task of disentanglement and introduce a new state-of-the-art approach called Non-synergistic variational Autoencoder.Our model draws inspiration from population coding, where the notion of synergy arises when we describe the encoded information by neurons in the form of responses from the stimuli.If those responses convey more information together than separate as independent sources of encoding information, they are acting synergetically.By penalizing the synergistic mutual information within the latents we encourage information independence and by doing that disentangle the latent factors.Notably, our approach could be added to the VAE framework easily, where the new ELBO function is still a lower bound on the log likelihood.In addition, we qualitatively compare our model with Factor VAE and show that this one implicitly minimises the synergy of the latents.","Minimising the synergistic mutual information within the latents and the data for the task of disentanglement using the VAE framework.Proposes a new objective function for learning dientangled representations in a variational framework by minimizing the synergy of the information provided.The authors aim at training a VAE that has disentangled latent representations in a ""synergistically"" maximal way. This paper proposes a new approach to enforcing disentanglement in VAEs using a term that penalizes the synergistic mutual information between the latent variables."
493,LSH Microbatches for Stochastic Gradients:  Value in Rearrangement," Metric embeddings are immensely useful representations of associations between entities . Embeddings are learned by optimizing a loss objective of the general form of a sum over example associations.Typically, the optimization uses stochastic gradient updates over minibatches of examples that are arranged independently at random.In this work, we propose the use of through randomized of examples that are more likely to include similar ones.We make a principled argument for the properties of our arrangements that accelerate the training and present efficient algorithms to generate microbatches that respect the marginal distribution of training examples. Finally, we observe experimentally that our structured arrangements accelerate training by 3-20%.Structured arrangements emerge as a powerful and novel performance knob for SGD that is independent and complementary to other SGD hyperparameters and thus is a candidate for wide deployment.",Accelerating SGD by arranging examples differentlyThe paper presents a method for improving the convergence rate of Stochastic Gradient Descent for learning embeddings by grouping similar training samples together.Proposes a non-uniform sampling strategy to construct minibatches in SGD for the task of learning embeddings for object associations.
494,Enhance Word Representation for Out-of-Vocabulary on Ubuntu Dialogue Corpus,"Ubuntu dialogue corpus is the largest public available dialogue corpus to make it feasible to build end-to-enddeep neural network models directly from the conversation data.One challenge of Ubuntu dialogue corpus isthe large number of out-of-vocabulary words.In this paper we proposed an algorithm which combines the general pre-trained word embedding vectors with those generated on the task-specific training set to address this issue. ', ""We integrated character embedding into Chen et al's Enhanced LSTM method and used it to evaluate the effectiveness of our proposed method.For the task of next utterance selection, the proposed method has demonstrated a significant performance improvement against original ESIM and the new model has achieved state-of-the-art results on both Ubuntu dialogue corpus and Douban conversation corpus.In addition, we investigated the performance impact of end-of-utterance and end-of-turn token tags.",Combine information between pre-built word embedding and task-specific word representation to address out-of-vocabulary issueThis paper proposes an approach to improve the out-of-vocabulary embedding prediction for the task of modeling dialogue conversations with sizable gains over the baselines.Proposes combining external pretrained word embeddings and pretrained word embeddings on training data by keeping them as two views.Proposes method to extend the coverage of pre-trained word embeddings to deal with the OOV problem that arises when applying them to conversational datasets and applies new variants of LSTM-based model to the task of response-selection in dialogue modeling.
495,Learned optimizers that outperform on wall-clock and validation loss,"Deep learning has shown that learned functions can dramatically outperform hand-designed functions on perceptual tasks.Analogously, this suggests that learned update functions may similarly outperform current hand-designed optimizers, especially for specific tasks.However, learned optimizers are notoriously difficult to train and have yet to demonstrate wall-clock speedups over hand-designed optimizers, and thus are rarely used in practice.Typically, learned optimizers are trained by truncated backpropagation through an unrolled optimization process.The resulting gradients are either strongly biased or have exploding norm.In this work we propose a training scheme which overcomes both of these difficulties, by dynamically weighting two unbiased gradient estimators for a variational loss on optimizer performance.This allows us to train neural networks to perform optimization faster than well tuned first-order methods.Moreover, by training the optimizer against validation loss, as opposed to training loss, we are able to use it to train models which generalize better than those trained by first order methods.We demonstrate these results on problems where our learned optimizer trains convolutional networks in a fifth of the wall-clock time compared to tuned first-order methods, and with an improvement","We analyze problems when training learned optimizers, address those problems via variational optimization using two complementary gradient estimators, and train optimizers that are 5x faster in wall-clock time than baseline optimizers (e.g. Adam).This paper uses un-rolled optimization to learn neural networks for optimization.This paper tackles the problem of learning an optimizer, specifically the authors focus on obtaining cleaner gradients from the unrolled training procedure.Presents a method for ""learning an optimizer"" by using a Variational Optimization for the ""outer"" optimizer loss and proposes the idea to combine both the reparametrized gradient and the score-function estimator for the Variational Objective and weights them using a product of Gaussians formula for the mean."
496,Asynchronous SGD without gradient delay for efficient distributed training,"Asynchronous distributed gradient descent algorithms for training of deep neuralnetworks are usually considered as inefficient, mainly because of the Gradient delayproblem.In this paper, we propose a novel asynchronous distributed algorithmthat tackles this limitation by well-thought-out averaging of model updates, computedby workers.The algorithm allows computing gradients along the processof gradient merge, thus, reducing or even completely eliminating worker idle timedue to communication overhead, which is a pitfall of existing asynchronous methods.We provide theoretical analysis of the proposed asynchronous algorithm,and show its regret bounds.According to our analysis, the crucial parameter forkeeping high convergence rate is the maximal discrepancy between local parametervectors of any pair of workers.As long as it is kept relatively small, theconvergence rate of the algorithm is shown to be the same as the one of a sequentialonline learning.Furthermore, in our algorithm, this discrepancy is boundedby an expression that involves the staleness parameter of the algorithm, and isindependent on the number of workers.This is the main differentiator betweenour approach and other solutions, such as Elastic Asynchronous SGD or DownpourSGD, in which that maximal discrepancy is bounded by an expression thatdepends on the number of workers, due to gradient delay problem.To demonstrateeffectiveness of our approach, we conduct a series of experiments on imageclassification task on a cluster with 4 machines, equipped with a commodity communicationswitch and with a single GPU card per machine.Our experimentsshow a linear scaling on 4-machine cluster without sacrificing the test accuracy,while eliminating almost completely worker idle time.Since our method allowsusing commodity communication switch, it paves a way for large scale distributedtraining performed on commodity clusters.",A method for an efficient asynchronous distributed training of deep learning models along with theoretical regret bounds.The paper proposes an algorithm to restrict the staleness in asynchronous SGD and provides theoretical analysisProposes a hybrid-algorithm to eliminate the gradient delay of asynchronous methods.
497,Defensive Quantization: When Efficiency Meets Robustness,"Neural network quantization is becoming an industry standard to efficiently deploy deep learning models on hardware platforms, such as CPU, GPU, TPU, and FPGAs.However, we observe that the conventional quantization approaches are vulnerable to adversarial attacks.This paper aims to raise people's awareness about the security of the quantized models, and we designed a novel quantization methodology to jointly optimize the efficiency and robustness of deep learning models.We first conduct an empirical study to show that vanilla quantization suffers more from adversarial attacks.We observe that the inferior robustness comes from the error amplification effect, where the quantization operation further enlarges the distance caused by amplified noise.Then we propose a novel Defensive Quantization method by controlling the Lipschitz constant of the network during quantization, such that the magnitude of the adversarial noise remains non-expansive during inference.Extensive experiments on CIFAR-10 and SVHN datasets demonstrate that our new quantization method can defend neural networks against adversarial examples, and even achieves superior robustness than their full-precision counterparts, while maintaining the same hardware efficiency as vanilla quantization approaches.As a by-product, DQ can also improve the accuracy of quantized models without adversarial attack.",We designed a novel quantization methodology to jointly optimize the efficiency and robustness of deep learning models. Proposes a regularization scheme to protect quantized neural networks from adversarial attacks using a Lipschitz constant filitering of the inner layers' inpout-output.
498,Skip RNN: Learning to Skip State Updates in Recurrent Neural Networks,"Recurrent Neural Networks continue to show outstanding performance in sequence modeling tasks.However, training RNNs on long sequences often face challenges like slow inference, vanishing gradients and difficulty in capturing long term dependencies.In backpropagation through time settings, these issues are tightly coupled with the large, sequential computational graph resulting from unfolding the RNN in time.We introduce the Skip RNN model which extends existing RNN models by learning to skip state updates and shortens the effective size of the computational graph.This model can also be encouraged to perform fewer state updates through a budget constraint.We evaluate the proposed model on various tasks and show how it can reduce the number of required RNN updates while preserving, and sometimes even improving, the performance of the baseline RNN models.Source code is publicly available at https://imatge-upc.github.io/skiprnn-2017-telecombcn/.","A modification for existing RNN architectures which allows them to skip state updates while preserving the performance of the original architectures.Proposes the Skip RNN model which allows a recurrent network to selectively skip updating its hidden state for some inputs, leading to reduced computation at test-time.Proposes a novel RNN model where both the input and the state update of the recurrent cells are skipped adaptively for some time steps."
499,Small steps and giant leaps: Minimal Newton solvers for Deep Learning,"We propose a fast second-order method that can be used as a drop-in replacement for current deep learning solvers.Compared to stochastic gradient descent, it only requires two additional forward-mode automatic differentiation operations per iteration, which has a computational cost comparable to two standard forward passes and is easy to implement.Our method addresses long-standing issues with current second-order solvers, which invert an approximate Hessian matrix every iteration exactly or by conjugate-gradient methods, procedures that are much slower than a SGD step.Instead, we propose to keep a single estimate of the gradient projected by the inverse Hessian matrix, and update it once per iteration with just two passes over the network.This estimate has the same size and is similar to the momentum variable that is commonly used in SGD.No estimate of the Hessian is maintained.We first validate our method, called CurveBall, on small problems with known solutions, where current deep learning solvers struggle.We then train several large models on CIFAR and ImageNet, including ResNet and VGG-f networks, where we demonstrate faster convergence with no hyperparameter tuning.We also show our optimiser's generality by testing on a large set of randomly-generated architectures.","A fast second-order solver for deep learning that works on ImageNet-scale problems with no hyper-parameter tuningChoosing direction by using a single step of gradient descent ""towards Newton step"" from an original estimate, and then taking this direction instead of original gradientA new approximate second-order optimization method with low computational cost that replaces the computation of the Hessian matrix with a single gradient step and a warm start strategy."
500,"Attention, Learn to Solve Routing Problems!","The recently presented idea to learn heuristics for combinatorial optimization problems is promising as it can save costly development.However, to push this idea towards practical implementation, we need better models and better ways of training.We contribute in both directions: we propose a model based on attention layers with benefits over the Pointer Network and we show how to train this model using REINFORCE with a simple baseline based on a deterministic greedy rollout, which we find is more efficient than using a value function.We significantly improve over recent learned heuristics for the Travelling Salesman Problem, getting close to optimal results for problems up to 100 nodes.With the same hyperparameters, we learn strong heuristics for two variants of the Vehicle Routing Problem, the Orienteering Problem and the Prize Collecting TSP, outperforming a wide range of baselines and getting results close to highly optimized and specialized algorithms.",Attention based model trained with REINFORCE with greedy rollout baseline to learn heuristics with competitive results on TSP and other routing problemsPresents an attention-based approach to learning a policy for solving TSP and other routing-type combinatorial optimzation problems.This paper trys to learn heuristics for solving combinatorial optimisation problems
501,On the State of the Art of Evaluation in Neural Language Models,"Ongoing innovations in recurrent neural network architectures have provided a steady influx of apparently state-of-the-art results on language modelling benchmarks.However, these have been evaluated using differing codebases and limited computational resources, which represent uncontrolled sources of experimental variation.We reevaluate several popular architectures and regularisation methods with large-scale automatic black-box hyperparameter tuning and arrive at the somewhat surprising conclusion that standard LSTM architectures, when properly regularised, outperform more recent models.We establish a new state of the art on the Penn Treebank and Wikitext-2 corpora, as well as strong baselines on the Hutter Prize dataset.","Show that LSTMs are as good or better than recent innovations for LM and that model evaluation is often unreliable.This paper describes a comprehensive validation of LSTM-based word and character language models, leading to a significant result in language modeling and a milestone in deep learning."
502,Investigating the effect of residual and highway connections in speech enhancement models," Residual and skip connections play an important role in many current generative models.Although their theoretical and numerical advantages are understood, their role in speech enhancement systems has not been investigated so far.When performing spectral speech enhancement, residual connections are very similar in nature to spectral subtraction, which is the one of the most commonly employed speech enhancement approaches. Highway networks, on the other hand, can be seen as a combination of spectral masking and spectral subtraction.However, when using deep neural networks, such operations would normally happen in a transformed spectral domain, as opposed to traditional speech enhancement where all operations are often done directly on the spectrum. In this paper, we aim to investigate the role of residual and highway connections in deep neural networks for speech enhancement, and verify whether or not they operate similarly to their traditional, digital signal processing counterparts.We visualize the outputs of such connections, projected back to the spectral domain, in models trained for speech denoising, and show that while skip connections do not necessarily improve performance with regards to the number of parameters, they make speech enhancement models more interpretable.","We show how using skip connections can make speech enhancement models more interpretable, as it makes them use similar mechanisms that have been explored in the DSP literature.The authors propose incorporating Residual, Highway and Masking blocks inside a fully convolutional pipeline in order to understand how iterative inference of the output and the masking is performed in a speech enhancement taskThe authors interpret highway, residual and masking connections. The authors generate their own noisy speech by artificially adding noise from a well established noise data-set to a less know clean speech data-set."
503,Deterministic Variational Inference for Robust Bayesian Neural Networks,"Bayesian neural networks hold great promise as a flexible and principled solution to deal with uncertainty when learning from finite data.Among approaches to realize probabilistic inference in deep neural networks, variational Bayes is theoretically grounded, generally applicable, and computationally efficient.With wide recognition of potential advantages, why is it that variational Bayes has seen very limited practical use for BNNs in real applications?We argue that variational inference in neural networks is fragile: successful implementations require careful initialization and tuning of prior variances, as well as controlling the variance of Monte Carlo gradient estimates.We provide two innovations that aim to turn VB into a robust inference tool for Bayesian neural networks: first, we introduce a novel deterministic method to approximate moments in neural networks, eliminating gradient variance; second, we introduce a hierarchical prior for parameters and a novel Empirical Bayes procedure for automatically selecting prior variances.Combining these two innovations, the resulting method is highly efficient and robust.On the application of heteroscedastic regression we demonstrate good predictive performance over alternative approaches.",A method for eliminating gradient variance and automatically tuning priors for effective training of bayesian neural networksProposes a new approach to perform deterministic variational inference for feed-forward BNN with specific nonlinear activation functions by approximating layerwise moments.The paper considers a purely deterministic approach to learning variational posterior approximations for Bayesian neural networks.
504,Automata Guided Skill Composition,Skills learned through reinforcement learning often generalizes poorlyacross tasks and re-training is necessary when presented with a new task.Wepresent a framework that combines techniques in formal methods with reinforcementlearning that allows for the convenient specification of complex temporaldependent tasks with logical expressions and construction of new skills from existingones with no additional exploration.We provide theoretical results for ourcomposition technique and evaluate on a simple grid world simulation as well asa robotic manipulation task.,"A formal method's approach to skill composition in reinforcement learning tasks"", 'The paper combines RL and constraints expressed by logical formulas by setting up an automation from scTLTL formulas.Proposes a method that helps to construct policy from learned subtasks on the topic of combining RL tasks with linear temporal logic formulas."
505,Deep Generative Models for learning Coherent Latent Representations from Multi-Modal Data,"The application of multi-modal generative models by means of a Variational Auto Encoder is an upcoming research topic for sensor fusion and bi-directional modality exchange.This contribution gives insights into the learned joint latent representation and shows that expressiveness and coherence are decisive properties for multi-modal datasets.Furthermore, we propose a multi-modal VAE derived from the full joint marginal log-likelihood that is able to learn the most meaningful representation for ambiguous observations.Since the properties of multi-modal sensor setups are essential for our approach but hardly available, we also propose a technique to generate correlated datasets from uni-modal ones.","Deriving a general formulation of a multi-modal VAE from the joint marginal log-likelihood.Proposes a multi-modal VAE with a variational bound derived from the chain rule.This paper proposes an objective, M^2VAE, for multi-modal VAEs, which is supposed to learn a more meaningful latent space representation."
506,Auto-Encoding Sequential Monte Carlo,"We build on auto-encoding sequential Monte Carlo: a method for model and proposal learning based on maximizing the lower bound to the log marginal likelihood in a broad family of structured probabilistic models.Our approach relies on the efficiency of sequential Monte Carlo for performing inference in structured probabilistic models and the flexibility of deep neural networks to model complex conditional probability distributions.We develop additional theoretical insights and introduce a new training procedure which improves both model and proposal learning.We demonstrate that our approach provides a fast, easy-to-implement and scalable means for simultaneous model learning and proposal adaptation in deep generative models.","We build on auto-encoding sequential Monte Carlo, gain new theoretical insights and develop an improved training procedure based on those insights.The paper proposes a version of IWAE-style training that uses SMC instead of classical importance sampling.This work proposes auto-encoding sequential Monte Carlo (SMC), extending the VAE framework to a new Monte Carto objective based on SMC. "
507,Two-Timescale Networks for Nonlinear Value Function Approximation,"A key component for many reinforcement learning agents is to learn a value function, either for policy evaluation or control.Many of the algorithms for learning values, however, are designed for linear function approximation with a fixed basis or fixed representation.Though there have been a few sound extensions to nonlinear function approximation, such as nonlinear gradient temporal difference learning, these methods have largely not been adopted, eschewed in favour of simpler but not sound methods like temporal difference learning and Q-learning.In this work, we provide a two-timescale network architecture that enables linear methods to be used to learn values, with a nonlinear representation learned at a slower timescale.The approach facilitates the use of algorithms developed for the linear setting, such as data-efficient least-squares methods, eligibility traces and the myriad of recently developed linear policy evaluation algorithms, to provide nonlinear value estimates.We prove convergence for TTNs, with particular care given to ensure convergence of the fast linear component under potentially dependent features provided by the learned representation.We empirically demonstrate the benefits of TTNs, compared to other nonlinear value function approximation algorithms, both for policy evaluation and control. ",We propose an architecture for learning value functions which allows the use of any linear policy evaluation algorithm in tandem with nonlinear feature learning.The paper proposes a two-timescale framework for learning the value function and a state representation altogether with nonlinear approximators.This paper proposes Two-Timescale Networks (TTNs) and prove the convergence of this method using methods from two time-scale stochastic approximation. This paper presents a Two-Timescale Network (TTN) that enables linear methods to be used to learn values. 
508,Low-Rank Matrix Factorization of LSTM as Effective Model Compression,"Large-scale Long Short-Term Memory cells are often the building blocks of many state-of-the-art algorithms for tasks in Natural Language Processing.However, LSTMs are known to be computationally inefficient because the memory capacity of the models depends on the number of parameters, and the inherent recurrence that models the temporal dependency is not parallelizable.In this paper, we propose simple, but effective, low-rank matrix factorization algorithms to compress network parameters and significantly speed up LSTMs with almost no loss of performance.To show the effectiveness of our method across different tasks, we examine two settings:1) compressing core LSTM layers in Language Models,2) compressing biLSTM layers of ELMo~p and evaluate in three downstream NLP tasks.The latter is particularly interesting as embeddings from large pre-trained biLSTM Language Models are often used as contextual word representations.Finally, we discover that matrix factorization performs better in general, additive recurrence is often more important than multiplicative recurrence, and we identify an interesting correlation between matrix norms and compression performance.","We propose simple, but effective, low-rank matrix factorization (MF) algorithms to speed up in running time, save memory, and improve the performance of LSTMs.Proposes to accelerate LSTM by using MF as the post-processing compression strategy and conducts extensive experiements to show the performance."
509,"A Forensic Representation to Detect Non-Trivial Image Duplicates, and How it Applies to Semantic Segmentation","Manipulation and re-use of images in scientific publications is a recurring problem, at present lacking a scalable solution. Existing tools for detecting image duplication are mostly manual or semi-automated, despite the fact that generating data for a learning-based approach is straightforward, as we here illustrate.This paper addresses the problem of determining if, given two images, one is a manipulated version of the other by means of certain geometric and statistical manipulations, e.g. copy, rotation, translation, scale, perspective transform, histogram adjustment, partial erasing, and compression artifacts.We propose a solution based on a 3-branch Siamese Convolutional Neural Network.The ConvNet model is trained to map images into a 128-dimensional space, where the Euclidean distance between duplicate images is no greater than 1.Our results suggest that such an approach can serve as tool to improve surveillance of the published and in-peer-review literature for image manipulation.We also show that as a byproduct the network learns useful representations for semantic segmentation, with performance comparable to that of domain-specific models.","A forensic metric to determine if a given image is a copy (with possible manipulation) of another image from a given dataset.Introduces the siamese network to identify duplicate and copied/modified images, which can be used to improve surveillance of the published and in-peer-review literature.The paper presents an application of deep convolutional networks for the task of duplicate image detectionThis work addresses the problem of finding duplicate/near duplicate images from biomedical publications and proposes a standard CNN and loss functions and apply it to this field."
510,Stabilizing GAN Training with Multiple Random Projections,"Training generative adversarial networks is unstable in high-dimensions as the true data distribution tends to be concentrated in a small fraction of the ambient space.The discriminator is then quickly able to classify nearly all generated samples as fake, leaving the generator without meaningful gradients and causing it to deteriorate after a point in training.In this work, we propose training a single generator simultaneously against an array of discriminators, each of which looks at a different random low-dimensional projection of the data.Individual discriminators, now provided with restricted views of the input, are unable to reject generated samples perfectly and continue to provide meaningful gradients to the generator throughout training. Meanwhile, the generator learns to produce samples consistent with the full data distribution to satisfy all discriminators simultaneously.We demonstrate the practical utility of this approach experimentally, and show that it is able to produce image samples with higher quality than traditional training with a single discriminator.","Stable GAN training in high dimensions by using an array of discriminators, each with a low dimensional view of generated samplesThe paper proposes to stabilize GAN training by using an ensemble of discriminators, each working on a random projection of the input data, to provide the training signal for the generator model.The paper proposes a GAN training method for improving the training stability. The paper proposes a new approach to GAN training, which provides stable gradients to train the generator."
511,Imposing Category Trees Onto Word-Embeddings Using A Geometric Construction,"We present a novel method to precisely impose tree-structured category information onto word-embeddings, resulting in ball embeddings in higher dimensional spaces.Inclusion relations among N-balls implicitly encode subordinate relations among categories.The similarity measurement in terms of the cosine function is enriched by category information.Using a geometric construction method instead of back-propagation, we create large N-ball embeddings that satisfy two conditions: category trees are precisely imposed onto word embeddings at zero energy cost; pre-trained word embeddings are well preserved.A new benchmark data set is created for validating the category of unknown words.Experiments show that N-ball embeddings, carrying category information, significantly outperform word embeddings in the test of nearest neighborhoods, and demonstrate surprisingly good performance in validating categories of unknown words.Source codes and data-sets are free for public access and .","we show a geometric method to perfectly encode categroy tree information into pre-trained word-embeddings.The paper proposes N-ball embedding for taxonomic data where an N-ball is a pair of a centroid vector and the radius from the center.The paper presents a method for tweaking existing vector embeddings of categorical objects (such as words), to convert them to ball embeddings that follow hierarchies.Focuses on adjusting the pretrained word embeddings so that they respect the hypernymy/hyponymy relationship by appropriate n-ball encapsulation."
512,Convolutional CRFs for Semantic Segmentation,"For the challenging semantic image segmentation task the best performing modelshave traditionally combined the structured modelling capabilities of ConditionalRandom Fields with the feature extraction power of CNNs.In more recentworks however, CRF post-processing has fallen out of favour.We argue that thisis mainly due to the slow training and inference speeds of CRFs, as well as thedifficulty of learning the internal CRF parameters.To overcome both issues wepropose to add the assumption of conditional independence to the framework offully-connected CRFs.This allows us to reformulate the inference in terms ofconvolutions, which can be implemented highly efficiently on GPUs.Doing sospeeds up inference and training by two orders of magnitude.All parameters ofthe convolutional CRFs can easily be optimized using backpropagation.Towardsthe goal of facilitating further CRF research we have made our implementationspublicly available.","We propose Convolutional CRFs a fast, powerful and trainable alternative to Fully Connected CRFs.The authors replace the large filtering step in the permutohedral lattice with a spatially varying convolutional kernel and show that inference is more efficient and training is easier. Proposes to perform message passing on a truncated Gaussian kernel CRF using a defined kernel and parallelized message passing on GPU."
513,Are you tough enough? Framework for Robustness Validation of Machine Comprehension Systems,"Deep Learning NLP domain lacks procedures for the analysis of model robustness.In this paper we propose a framework which validates robustness of any Question Answering model through model explainers.We propose that output of a robust model should be invariant to alterations that do not change its semantics.We test this property by manipulating question in two ways: swapping important question word for1) its semantically correct synonym and2) for word vector that is close in embedding space.We estimate importance of words in asked questions with Locally Interpretable Model Agnostic Explanations method.With these two steps we compare state-of-the-art Q&A models.We show that although accuracy of state-of-the-art models is high, they are very fragile to changes in the input.We can choose architecture that is more immune to attacks and thus more robust and stable in production environment.Morevoer, we propose 2 adversarial training scenarios which raise model sensitivity to true synonyms by up to 7% accuracy measure.Our findings help to understand which models are more stable and how they can be improved.In addition, we have created and published a new dataset that may be used for validation of robustness of a Q&A model.",We propose a model agnostic approach to validation of Q&A system robustness and demonstrate results on state-of-the-art Q&A models.Addresses the problem of robustness to adversarial information in question answering.Improving robustness of machine comprehension/question answering.
514,Counterfactual Image Networks,"We capitalize on the natural compositional structure of images in order to learn object segmentation with weakly labeled images.The intuition behind our approach is that removing objects from images will yield natural images, however removing random patches will yield unnatural images.We leverage this signal to develop a generative model that decomposes an image into layers, and when all layers are combined, it reconstructs the input image.However, when a layer is removed, the model learns to produce a different image that still looks natural to an adversary, which is possible by removing objects.Experiments and visualizations suggest that this model automatically learns object segmentation on images labeled only by scene better than baselines.","Weakly-supervised image segmentation using compositional structure of images and generative models.This paper creates a layered representation in order to better learn segmentation from unlabeled images.This paper proposes a GAN-based generative model that decomposes images into multiple layers, where the objective of the GAN is to distinguish real images from images formed by combining the layers.This paper proposes a neural network architecture around the idea of layered scene composition"
515,On the Geometry of Adversarial Examples,"Adversarial examples are a pervasive phenomenon of machine learning models where seemingly imperceptible perturbations to the input lead to misclassifications for otherwise statistically accurate models.We propose a geometric framework, drawing on tools from the manifold reconstruction literature, to analyze the high-dimensional geometry of adversarial examples.In particular, we highlight the importance of codimension: for low-dimensional data manifolds embedded in high-dimensional space there are many directions off the manifold in which to construct adversarial examples.Adversarial examples are a natural consequence of learning a decision boundary that classifies the low-dimensional data manifold well, but classifies points near the manifold incorrectly.Using our geometric framework we prove a tradeoff between robustness under different norms, that adversarial training in balls around the data is sample inefficient, and sufficient sampling conditions under which nearest neighbor classifiers and ball-based adversarial training are robust.","We present a geometric framework for proving robustness guarantees and highlight the importance of codimension in adversarial examples. This paper gives a theoretical analysis of adversarial examples, showing that there exists a tradeoff between robustness in different norms, adversarial training is sample inefficient, and the nearest neighbor classifier can be robust under certain conditions."
516,Synthetic and Natural Noise Both Break Neural Machine Translation,"Character-based neural machine translation models alleviate out-of-vocabulary issues, learn morphology, and move us closer to completely end-to-end translation systems. Unfortunately, they are also very brittle and easily falter when presented with noisy data.In this paper, we confront NMT models with synthetic and natural sources of noise.We find that state-of-the-art models fail to translate even moderately noisy texts that humans have no trouble comprehending.We explore two approaches to increase model robustness: structure-invariant word representations and robust training on noisy texts.We find that a model based on a character convolutional neural network is able to simultaneously learn representations robust to multiple kinds of noise.","CharNMT is brittleThis paper investigates the impact of character-level noise on 4 different neural machine translation systemsThis paper empirically investigates the performance of character-level NMT systems in the face of character-level noise, both synthesized and natural.This paper investigates the impact of noisy input on Machine Translation and tests ways to make NMT models more robust"
517,Deep Learning as a Mixed Convex-Combinatorial Optimization Problem,"As neural networks grow deeper and wider, learning networks with hard-threshold activations is becoming increasingly important, both for network quantization, which can drastically reduce time and energy requirements, and for creating large integrated systems of deep networks, which may have non-differentiable components and must avoid vanishing and exploding gradients for effective learning.However, since gradient descent is not applicable to hard-threshold functions, it is not clear how to learn them in a principled way.We address this problem by observing that setting targets for hard-threshold hidden units in order to minimize loss is a discrete optimization problem, and can be solved as such.The discrete optimization goal is to find a set of targets such that each unit, including the output, has a linearly separable problem to solve.Given these targets, the network decomposes into individual perceptrons, which can then be learned with standard convex approaches.Based on this, we develop a recursive mini-batch algorithm for learning deep hard-threshold networks that includes the popular but poorly justified straight-through estimator as a special case.Empirically, we show that our algorithm improves classification accuracy in a number of settings, including for AlexNet and ResNet-18 on ImageNet, when compared to the straight-through estimator.","We learn deep networks of hard-threshold units by setting hidden-unit targets using combinatorial optimization and weights by convex optimization, resulting in improved performance on ImageNet.The paper explains and generalizes approaches for learning neural nets with hard activation.This paper examines the problem of optimizing deep networks of hard-threshold units.The paper discusses the problem of optimizing neural networks with hard threshold and proposes a novel solution to it with a collection of heuristics/approximations."
518,Not-So-CLEVR: Visual Relations Strain Feedforward Neural Networks,"The robust and efficient recognition of visual relations in images is a hallmark of biological vision.Here, we argue that, despite recent progress in visual recognition, modern machine vision algorithms are severely limited in their ability to learn visual relations.Through controlled experiments, we demonstrate that visual-relation problems strain convolutional neural networks.The networks eventually break altogether when rote memorization becomes impossible such as when the intra-class variability exceeds their capacity.We further show that another type of feedforward network, called a relational network, which was shown to successfully solve seemingly difficult visual question answering problems on the CLEVR datasets, suffers similar limitations.Motivated by the comparable success of biological vision, we argue that feedback mechanisms including working memory and attention are the key computational components underlying abstract visual reasoning.","Using a novel, controlled, visual-relation challenge, we show that same-different tasks critically strain the capacity of CNNs; we argue that visual relations can be better solved using attention-mnemonic strategies.Demonstrates that convolutional and relational neural networks fail to solve visual relation problems by training networks on artificially generated visual relation data. This paper explores how current CNN's and Relational Networks fail to recognize visual relations in images."
519,AD-VAT: An Asymmetric Dueling mechanism for learning Visual Active Tracking,"Visual Active Tracking aims at following a target object by autonomously controlling the motion system of a tracker given visual observations.Previous work has shown that the tracker can be trained in a simulator via reinforcement learning and deployed in real-world scenarios.However, during training, such a method requires manually specifying the moving path of the target object to be tracked, which cannot ensure the tracker’s generalization on the unseen object moving patterns.To learn a robust tracker for VAT, in this paper, we propose a novel adversarial RL method which adopts an Asymmetric Dueling mechanism, referred to as AD-VAT.In AD-VAT, both the tracker and the target are approximated by end-to-end neural networks, and are trained via RL in a dueling/competitive manner: i.e., the tracker intends to lockup the target, while the target tries to escape from the tracker.They are asymmetric in that the target is aware of the tracker, but not vice versa.Specifically, besides its own observation, the target is fed with the tracker’s observation and action, and learns to predict the tracker’s reward as an auxiliary task.We show that such an asymmetric dueling mechanism produces a stronger target, which in turn induces a more robust tracker.To stabilize the training, we also propose a novel partial zero-sum reward for the tracker/target.The experimental results, in both 2D and 3D environments, demonstrate that the proposed method leads to a faster convergence in training and yields more robust tracking behaviors in different testing scenarios.For supplementary videos, see: https://www.youtube.com/playlist?list=PL9rZj4Mea7wOZkdajK1TsprRg8iUf51BS The code is available at https://github.com/zfw1226/active_tracking_rl","We propose AD-VAT, where the tracker and the target object, viewed as two learnable agents, are opponents and can mutually enhance during training.This work aims to address the visual active tracking problem with a training mechanism in which the tracker and target serve as mutual opponentsThis paper presents a simple multi-agent Deep RL task where a moving tracker tries to follow a moving target.Proposes a novel reward function - ""partial zero sum"", which only encourages the tracker-target competition when they are close and penalizes whey they are too far."
520,Joint Learning of Hierarchical Word Embeddings from a Corpus and a Taxonomy,"Identifying the hypernym relations that hold between words is a fundamental task in NLP.Word embedding methods have recently shown some capability to encode hypernymy.However, such methods tend not to explicitly encode the hypernym hierarchy that exists between words.In this paper, we propose a method to learn a hierarchical word embedding in a specific order to capture the hypernymy.To learn the word embeddings, the proposed method considers not only the hypernym relations that exists between words on a taxonomy, but also their contextual information in a large text corpus.The experimental results on a supervised hypernymy detection and a newly-proposed hierarchical path completion tasks show the ability of the proposed method to encode the hierarchy.Moreover, the proposed method outperforms previously proposed methods for learning word and hypernym-specific word embeddings on multiple benchmarks.","We presented a method to jointly learn a Hierarchical Word Embedding (HWE) using a corpus and a taxonomy for identifying the hypernymy relations between words.The paper presents a method to jointly learn word embeddings using co-occurrence statistics as well as incorporating hierarchal information from semantic networks.This paper proposed a joint learning method of hypernym from both raw text and supervised taxonomy data. This paper proposes adding a measure of ""distributional inclusion"" difference to the GloVE objective for the purpose of representing hypernym relations."
521,Precision Highway for Ultra Low-precision Quantization,"Quantization of a neural network has an inherent problem called accumulated quantization error, which is the key obstacle towards ultra-low precision, e.g., 2- or 3-bit precision.To resolve this problem, we propose precision highway, which forms an end-to-end high-precision information flow while performing the ultra-low-precision computation.First, we describe how the precision highway reduce the accumulated quantization error in both convolutional and recurrent neural networks.We also provide the quantitative analysis of the benefit of precision highway and evaluate the overhead on the state-of-the-art hardware accelerator.In the experiments, our proposed method outperforms the best existing quantization methods while offering 3-bit weight/activation quantization with no accuracy loss and 2-bit quantization with a 2.45 % top-1 accuracy loss in ResNet-50.We also report that the proposed method significantly outperforms the existing method in the 2-bit quantization of an LSTM for language modeling.","precision highway; a generalized concept of high-precision information flow for sub 4-bit quantization Investigates the problem of neural network quantization by employing an end-to-end precision highway to reduce the accumulated quantization error and enable ultra-low precision in deep neural networks. This paper studies methods to improve the performance of quantized neural networksThis paper proposes to keep a high activation/gradient flow in two kinds of networks structures, ResNet and LSTM."
522,Temporally Efficient Deep Learning with Spikes,"The vast majority of natural sensory data is temporally redundant.For instance, video frames or audio samples which are sampled at nearby points in time tend to have similar values. Typically, deep learning algorithms take no advantage of this redundancy to reduce computations. This can be an obscene waste of energy. We present a variant on backpropagation for neural networks in which computation scales with the rate of change of the data - not the rate at which we process the data. We do this by implementing a form of Predictive Coding wherein neurons communicate a combination of their state, and their temporal change in state, and quantize this signal using Sigma-Delta modulation. Intriguingly, this simple communication rule give rise to units that resemble biologically-inspired leaky integrate-and-fire neurons, and to a spike-timing-dependent weight-update similar to Spike-Timing Dependent Plasticity, a synaptic learning rule observed in the brain. We demonstrate that on MNIST, on a temporal variant of MNIST, and on Youtube-BB, a dataset with videos in the wild, our algorithm performs about as well as a standard deep network trained with backpropagation, despite only communicating discrete values between layers. ","An algorithm for training neural networks efficiently on temporally redundant data.The paper describes a neural coding scheme for spike based learning in deep neural networksThis paper presents a method for spike based learning that aims at reducing the needed computation during learning and testing when classifying temporal redundant data.This paper applies a predictive coding version of the Sigma-Delta encoding scheme to reduce a computational load on a deep learning network, combining the three components in a way not seen previously."
523,Caveats for information bottleneck in deterministic scenarios,"Information bottleneck is a method for extracting information from one random variable X that is relevant for predicting another random variable Y. To do so, IB identifies an intermediate ""bottleneck"" variable T that has low mutual information I and high mutual information I.The ""IB curve"" characterizes the set of bottleneck variables that achieve maximal I for a given I, and is typically explored by maximizing the ""IB Lagrangian"", I - βI.In some cases, Y is a deterministic function of X, including many classification problems in supervised learning where the output class Y is a deterministic function of the input X. We demonstrate three caveats when using IB in any situation where Y is a deterministic function of X: the IB curve cannot be recovered by maximizing the IB Lagrangian for different values of β; there are ""uninteresting"" trivial solutions at all points of the IB curve; and for multi-layer classifiers that achieve low prediction error, different layers cannot exhibit a strict trade-off between compression and prediction, contrary to a recent proposal.We also show that when Y is a small perturbation away from being a deterministic function of X, these three caveats arise in an approximate way.To address problem, we propose a functional that, unlike the IB Lagrangian, can recover the IB curve in all cases.We demonstrate the three caveats on the MNIST dataset.",Information bottleneck behaves in surprising ways whenever the output is a deterministic function of the input.Argues that most real classification problems show such a deterministic relation between the class labels and the inputs X and explores several issues that result from such pathologies.Explores issues that arise when applying information bottlenext concepts to deterministic supervised learning modelsThe authors clarify several counter-intuitive behaviors of the information bottleneck method for supervised learning of a deterministic rule.
524,Sufficient Conditions for Robustness to Adversarial Examples: a Theoretical and Empirical Study with Bayesian Neural Networks,"We prove, under two sufficient conditions, that idealised models can have no adversarial examples.We discuss which idealised models satisfy our conditions, and show that idealised Bayesian neural networks satisfy these.We continue by studying near-idealised BNNs using HMC inference, demonstrating the theoretical ideas in practice.We experiment with HMC on synthetic data derived from MNIST for which we know the ground-truth image density, showing that near-perfect epistemic uncertainty correlates to density under image manifold, and that adversarial images lie off the manifold in our setting.This suggests why MC dropout, which can be seen as performing approximate inference, has been observed to be an effective defence against adversarial examples in practice; We highlight failure-cases of non-idealised BNNs relying on dropout, suggesting a new attack for dropout models and a new defence as well.Lastly, we demonstrate the defence on a cats-vs-dogs image classification task with a VGG13 variant.","We prove that idealised Bayesian neural networks can have no adversarial examples, and give empirical evidence with real-world BNNs.The paper studies the adversarial robustness of Bayesian classifiers and state two conditions that they show are provably sufficient for ""idealised models"" on ""idealised datasets"" to not have adversarial examplesPaper posit a class of discriminative BAyesian classifiers that do not have any adversarial examples."
525,Adversarial Reprogramming of Neural Networks,"Deep neural networks are susceptible to adversarial attacks.In computer vision, well-crafted perturbations to images can cause neural networks to make mistakes such as confusing a cat with a computer.Previous adversarial attacks have been designed to degrade performance of models or cause machine learning models to produce specific outputs chosen ahead of time by the attacker.We introduce attacks that instead reprogram the target model to perform a task chosen by the attacker without the attacker needing to specify or compute the desired output for each test-time input.This attack finds a single adversarial perturbation, that can be added to all test-time inputs to a machine learning model in order to cause the model to perform a task chosen by the adversary—even if the model was not trained to do this task.These perturbations can thus be considered a program for the new task.We demonstrate adversarial reprogramming on six ImageNet classification models, repurposing these models to perform a counting task, as well as classification tasks: classification of MNIST and CIFAR-10 examples presented as inputs to the ImageNet model.","We introduce the first instance of adversarial attacks that reprogram the target model to perform a task chosen by the attacker---without the attacker needing to specify or compute the desired output for each test-time input.The authors present a novel adversarial attack scheme where a neural net is repurposed to accomplish a different task than the one it was originally trained onThis paper proposed ""adversarial reprogramming"" of well-trained and fixed neural networks and show that adversarial reprogramming is less effective on untrained networks. The paper extends the idea of 'adversarial attacks' in supervised learning of NNs to a full repurposing of the solution of a trained net."
526,Predicting the Generalization Gap in Deep Networks with Margin Distributions,"As shown in recent research, deep neural networks can perfectly fit randomly labeled data, but with very poor accuracy on held out data.This phenomenon indicates that loss functions such as cross-entropy are not a reliable indicator of generalization.This leads to the crucial question of how generalization gap should be predicted from the training data and network parameters.In this paper, we propose such a measure, and conduct extensive empirical studies on how well it can predict the generalization gap.Our measure is based on the concept of margin distribution, which are the distances of training points to the decision boundary.We find that it is necessary to use margin distributions at multiple layers of a deep network.On the CIFAR-10 and the CIFAR-100 datasets, our proposed measure correlates very strongly with the generalization gap.In addition, we find the following other factors to be of importance: normalizing margin values for scale independence, using characterizations of margin distribution rather than just the margin, and working in log space instead of linear space.Our measure can be easily applied to feedforward deep networks with any architecture and may point towards new training loss functions that could enable better generalization.","We develop a new scheme to predict the generalization gap in deep networks with high accuracy.Authors suggest using a geometric margin and layer-wise margin distribution for predicting generalization gap.Empirically shows an interesting connection between the proposed margin statistics and the generalization gap, which can be used to provide some prescriptive insights towards understanding generalization in deep neural nets. "
527,Improved Learning of One-hidden-layer Convolutional Neural Networks with Overlaps,"We propose a new algorithm to learn a one-hidden-layer convolutional neural network where both the convolutional weights and the outputs weights are parameters to be learned.Our algorithm works for a general class of patches, including commonly used structures for computer vision tasks.Our algorithm draws ideas from isotonic regression for learning neural networks and landscape analysis of non-convex matrix factorization problems.We believe these findings may inspire further development in designing provable algorithms for learning neural networks and other complex models.While our focus is theoretical, we also present experiments that illustrate our theoretical findings.","We propose an algorithm for provably recovering parameters (convolutional and output weights) of a convolutional network with overlapping patches.This paper studies the theoretical learning of one-hidden-layer convolutional neural nets, resulting in a learning algorithm and provable guarantees using the algorithm.This paper gives a new algorithm for learning a two layer neural network which involves a single convolutional filter and a weight vector for different locations."
528,On the regularization of Wasserstein GANs,"Since their invention, generative adversarial networks have become a popular approach for learning to model a distribution of real data.Convergence problems during training are overcome by Wasserstein GANs which minimize the distance between the model and the empirical distribution in terms of a different metric, but thereby introduce a Lipschitz constraint into the optimization problem.A simple way to enforce the Lipschitz constraint on the class of functions, which can be modeled by the neural network, is weight clipping.Augmenting the loss by a regularization term that penalizes the deviation of the gradient norm of the critic from one, was proposed as an alternative that improves training.We present theoretical arguments why using a weaker regularization term enforcing the Lipschitz constraint is preferable.These arguments are supported by experimental results on several data sets.","A new regularization term can improve your training of wasserstein gansThe paper proposes a regularization scheme for Wasserstein GAN based on relaxation of the constraints on the Lipschitz constant of 1.The article deals with regularization/penalization in the fitting of GANs, when based on a L_1 Wasserstein metric."
529,A Weighted Mini-Bucket Bound Heuristic for Solving Influence Diagrams,"Influence diagrams provide a modeling and inference framework for sequential decision problems, representing the probabilistic knowledge by a Bayesian network and the preferences of an agent by utility functions over the random variables and decision variables.MDPs and POMDPS, widely used for planning under uncertainty can also be represented by influence diagrams.The time and space complexity of computing the maximum expected utility and its maximizing policy is exponential in the induced width of the underlying graphical model, which is often prohibitively large due to the growth of the information set under the sequence of decisions.In this paper, we develop a weighted mini-bucket approach for bounding the MEU. These bounds can be used as a stand-alone approximation that can be improved as a function of a controlling i-bound parameter.They can also be used as heuristic functions to guide search, especially for planningsuch as MDPs and POMDPs.We evaluate the scheme empirically against state-of-the-art, thus illustrating its potential.","This paper introduces an elimination based heuristic function for sequential decision making, suitable for guiding AND/OR search algorithms for solving influence diagrams.generalizes minibuckets inference heuristic to influence diagrams."
530,Feed-forward Propagation in Probabilistic Neural Networks with Categorical and Max Layers,"Probabilistic Neural Networks deal with various sources of stochasticity: input noise, dropout, stochastic neurons, parameter uncertainties modeled as random variables, etc.In this paper we revisit a feed-forward propagation approach that allows one to estimate for each neuron its mean and variance w.r.t. all mentioned sources of stochasticity.In contrast, standard NNs propagate only point estimates, discarding the uncertainty.Methods propagating also the variance have been proposed by several authors in different context.The view presented here attempts to clarify the assumptions and derivation behind such methods, relate them to classical NNs and broaden their scope of applicability.The main technical contributions are new approximations for the distributions of argmax and max-related transforms, which allow for fully analytic uncertainty propagation in networks with softmax and max-pooling layers as well as leaky ReLU activations.We evaluate the accuracy of the approximation and suggest a simple calibration.Applying the method to networks with dropout allows for faster training and gives improved test likelihoods without the need of sampling.","Approximating mean and variance of the NN output over noisy input / dropout / uncertain parameters. Analytic approximations for argmax, softmax and max layers.The authors focus on the problem of uncertainty propagation DNNThis paper revisits the feed-forward propagation of mean and variance in neurons, by addressing the problem of propagating uncertainty through max-pooling layers and softmax."
531,Don't let your Discriminator  be fooled,"Generative Adversarial Networks are one of the leading tools in generative modeling, image editing and content creation.However, they are hard to train as they require a delicate balancing act between two deep networks fighting a never ending duel.Some of the most promising adversarial models today minimize a Wasserstein objective.It is smoother and more stable to optimize.In this paper, we show that the Wasserstein distance is just one out of a large family of objective functions that yield these properties.By making the discriminator of a GAN robust to adversarial attacks we can turn any GAN objective into a smooth and stable loss.We experimentally show that any GAN objective, including Wasserstein GANs, benefit from adversarial robustness both quantitatively and qualitatively.The training additionally becomes more robust to suboptimal choices of hyperparameters, model architectures, or objective functions.","A discriminator that is not easily fooled by adversarial example makes GAN training more robust and leads to a smoother objective.This paper proposes a new way to stabilize the training process of GAN by regularizing the Discriminator to be robust to adversarial examples.The paper proposes a systematic way of training GANs with robustness regularization terms, allowing for smoother training of GANs. Presents idea that making a discriminator robust to adversarial perturbations the GAN objective can be made smooth which results in better results both visually and in terms of FID."
532,Gaussian Process Neurons,"We propose a method to learn stochastic activation functions for use in probabilistic neural networks.First, we develop a framework to embed stochastic activation functions based on Gaussian processes in probabilistic neural networks.Second, we analytically derive expressions for the propagation of means and covariances in such a network, thus allowing for an efficient implementation and training without the need for sampling.Third, we show how to apply variational Bayesian inference to regularize and efficiently train this model.The resulting model can deal with uncertain inputs and implicitly provides an estimate of the confidence of its predictions.Like a conventional neural network it can scale to datasets of arbitrary size and be extended with convolutional and recurrent connections, if desired.",We model the activation function of each neuron as a Gaussian Process and learn it alongside the weight with Variational Inference.Propose placing Gaussian process priors on the functional form of each activation function in the neural net to learn the form of activation functions.
533,The Expressive Power of Deep Neural Networks with Circulant Matrices,"Recent results from linear algebra stating that any matrix can be decomposed into products of diagonal and circulant matrices has lead to the design of compact deep neural network architectures that perform well in practice.In this paper, we bridge the gap between these good empirical resultsand the theoretical approximation capabilities of Deep diagonal-circulant ReLU networks.More precisely, we first demonstrate that a Deep diagonal-circulant ReLU networks ofbounded width and small depth can approximate a deep ReLU network in which the dense matrices areof low rank.Based on this result, we provide new bounds on the expressive power and universal approximativeness of this type of networks.We support our experimental results with thorough experiments on a large, real world video classification problem.",We provide a theoretical study of the properties of Deep circulant-diagonal ReLU Networks and demonstrate that they are bounded width universal approximators.The paper proposes using circulant and diagonal matrices to speed up computation and reduce memory requirements in eural networks.This paper proves that bounded width diagonal-circulent ReLU networks (DC-ReLU) are universal approximators.
534,StarHopper: A Touch Interface for Remote Object-Centric Drone Navigation,"Camera drones, a rapidly emerging technology, offer people the ability to remotely inspect an environment with a high degree of mobility and agility.However, manual remote piloting of a drone is prone to errors.In contrast, autopilot systems can require a significant degree of environmental knowledge and are not necessarily designed to support flexible visual inspections.Inspired by camera manipulation techniques in interactive graphics, we designed StarHopper, a novel touch screen interface for efficient object-centric camera drone navigation, in which a user directly specifies the navigation of a drone camera relative to a specified object of interest.The system relies on minimal environmental information and combines both manual and automated control mechanisms to give users the freedom to remotely explore an environment with efficiency and accuracy.A lab study shows that StarHopper offers an efficiency gain of 35.4% over manual piloting, complimented by an overall user preference towards our object-centric navigation system.","StarHopper is a novel touch screen interface for efficient and flexible object-centric camera drone navigationThe authors outline a new drone control interface StarHopper that they have developed, combining automated and manual piloting into a new hybrid navigation interface and gets rid of the assumption that the target object is already in the drone’s FOV by using an additional overhead camera.This paper presents StarHopper, a system for semi-automatic drone navigation in the context of remote inspection.Introduces StarHopper, an application that uses computer vision techniques with touch input to support drone piloting with an object-centric approach."
535,Bi-Directional Block Self-Attention for Fast and Memory-Efficient Sequence Modeling,"Recurrent neural networks, convolutional neural networks and self-attention networks are commonly used to produce context-aware representations.RNN can capture long-range dependency but is hard to parallelize and not time-efficient.CNN focuses on local dependency but does not perform well on some tasks.SAN can model both such dependencies via highly parallelizable computation, but memory requirement grows rapidly in line with sequence length.In this paper, we propose a model, called ""bi-directional block self-attention network"", for RNN/CNN-free sequence encoding.It requires as little memory as RNN but with all the merits of SAN.Bi-BloSAN splits the entire sequence into blocks, and applies an intra-block SAN to each block for modeling local context, then applies an inter-block SAN to the outputs for all blocks to capture long-range dependency.Thus, each SAN only needs to process a short sequence, and only a small amount of memory is required.Additionally, we use feature-level attention to handle the variation of contexts around the same word, and use forward/backward masks to encode temporal order information.On nine benchmark datasets for different NLP tasks, Bi-BloSAN achieves or improves upon state-of-the-art accuracy, and shows better efficiency-memory trade-off than existing RNN/CNN/SAN.","A self-attention network for RNN/CNN-free sequence encoding with small memory consumption, highly parallelizable computation and state-of-the-art performance on several NLP tasksProposes applyting self-attention at two levels to limit the memory requirement in attention-based models with a negligible impact on speed.This paper introduces bi-directional block self-attention model as a general-purpose encoder for various sequence modeling tasks in NLP"
536,Coarse-grain Fine-grain Coattention Network for Multi-evidence Question Answering,"End-to-end neural models have made significant progress in question answering, however recent studies show that these models implicitly assume that the answer and evidence appear close together in a single document.In this work, we propose the Coarse-grain Fine-grain Coattention Network, a new question answering model that combines information from evidence across multiple documents.The CFC consists of a coarse-grain module that interprets documents with respect to the query then finds a relevant answer, and a fine-grain module which scores each candidate answer by comparing its occurrences across all of the documents with the query.We design these modules using hierarchies of coattention and self-attention, which learn to emphasize different parts of the input.On the Qangaroo WikiHop multi-evidence question answering task, the CFC obtains a new state-of-the-art result of 70.6% on the blind test set, outperforming the previous best by 3% accuracy despite not using pretrained contextual encoders.",A new state-of-the-art model for multi-evidence question answering using coarse-grain fine-grain hierarchical attention.Proposes a method for multi-hop QA based on two separate modules (coarse-grained and fine-grained modules).This paper proposes an interesting coarse-grain fine-grain coattention network architecture to address multi-evidence question answeringFocuses on multi-choice QA and proposes a coarse-to-fine scoring framework.
537,Scalable Unbalanced Optimal Transport using Generative Adversarial Networks,"Generative adversarial networks are an expressive class of neural generative models with tremendous success in modeling high-dimensional continuous measures.In this paper, we present a scalable method for unbalanced optimal transport based on the generative-adversarial framework.We formulate unbalanced OT as a problem of simultaneously learning a transport map and a scaling factor that push a source measure to a target measure in a cost-optimal manner.We provide theoretical justification for this formulation, showing that it is closely related to an existing static formulation by Liero et al..We then propose an algorithm for solving this problem based on stochastic alternating gradient updates, similar in practice to GANs, and perform numerical experiments demonstrating how this methodology can be applied to population modeling.",We propose new methodology for unbalanced optimal transport using generative adversarial networks.The authors consider the unbalanced optimal transport problem between two measures with different total mass using a stochastic min-max algorithm and local scalingThe authors propose an approach to estimate unbalanced optimal transport between sampled measures that scales well in the dimension and in the number of samples.The paper introduces a static formulation for unbalanced optimal transport by learning simultaneously a transport map T and scaling factor xi.
538,Classifier-agnostic saliency map extraction,"Extracting saliency maps, which indicate parts of the image important to classification, requires many tricks to achieve satisfactory performance when using classifier-dependent methods.Instead, we propose classifier-agnostic saliency map extraction, which finds all parts of the image that any classifier could use, not just one given in advance.We observe that the proposed approach extracts higher quality saliency maps and outperforms existing weakly-supervised localization techniques, setting the new state of the art result on the ImageNet dataset.","We propose a new saliency map extraction method which results in extracting higher quality maps.Proposes a classifier-agnostic method for saliency map extraction.This paper introduces a new saliency map extractor that seems to improve state-of-the-art results.The authors argue that when an extracted saliency map is directly dependent on a model, then it might not be useful for a different classifier, and suggests a scheme to approximate the solution."
539,InstaGAN: Instance-aware Image-to-Image Translation,"Unsupervised image-to-image translation has gained considerable attention due to the recent impressive progress based on generative adversarial networks.However, previous methods often fail in challenging cases, in particular, when an image has multiple target instances and a translation task involves significant changes in shape, e.g., translating pants to skirts in fashion images.To tackle the issues, we propose a novel method, coined instance-aware GAN, that incorporates the instance information and improves multi-instance transfiguration.The proposed method translates both an image and the corresponding set of instance attributes while maintaining the permutation invariance property of the instances.To this end, we introduce a context preserving loss that encourages the network to learn the identity function outside of target instances.We also propose a sequential mini-batch inference/training technique that handles multiple instances with a limited GPU memory and enhances the network to generalize better for multiple instances.Our comparative evaluation demonstrates the effectiveness of the proposed method on different image datasets, in particular, in the aforementioned challenging cases.Code and results are available in https://github.com/sangwoomo/instagan","We propose a novel method to incorporate the set of instance attributes for image-to-image translation.This paper proposes a method -- InstaGAN -- which builds on CycleGAN by taking into account instance information in the form of per-instance segmentation masks, with results that compare favorably to CycleGAN and other baselines. Proposes to add instance-aware segmentation masks for the problem of unpaired image-to-image translation."
540,Deep learning generalizes because the parameter-function map is biased towards simple functions,"Deep neural networks generalize remarkably well without explicit regularization even in the strongly over-parametrized regime where classical learning theory would instead predict that they would severely overfit. While many proposals for some kind of implicit regularization have been made to rationalise this success, there is no consensus for the fundamental reason why DNNs do not strongly overfit. In this paper, we provide a new explanation.By applying a very general probability-complexity bound recently derived from algorithmic information theory, we argue that the parameter-function map of many DNNs should be exponentially biased towards simple functions.We then provide clear evidence for this strong simplicity bias in a model DNN for Boolean functions, as well as in much larger fully connected and convolutional networks trained on CIFAR10 and MNIST.As the target functions in many real problems are expected to be highly structured, this intrinsic simplicity bias helps explain why deep networks generalize well on real world problems.This picture also facilitates a novel PAC-Bayes approach where the prior is taken over the DNN input-output function space, rather than the more conventional prior over parameter space. If we assume that the training algorithm samples parameters close to uniformly within the zero-error region then the PAC-Bayes theorem can be used to guarantee good expected generalization for target functions producing high-likelihood training sets. By exploiting recently discovered connections between DNNs and Gaussian processes to estimate the marginal likelihood, we produce relatively tight generalization PAC-Bayes error bounds which correlate well with the true error on realistic datasets such as MNIST and CIFAR10 and for architectures including convolutional and fully connected networks.","The parameter-function map of deep networks is hugely biased; this can explain why they generalize. We use PAC-Bayes and Gaussian processes to obtain nonvacuous bounds.The paper studies the generalization capabilities of deep neural networks, with the help of the PAC-Bayesian learning theory and empirically backed intuitions.This paper proposes an explaination of the generalization behaviors of large over-parameterized neural networks by claiming the parameter-function map in neural networks are biased towards ""simple"" functions and generalization behavior will be good if the target concept is also ""simple""."
541,Evolutionary Expectation Maximization for Generative Models with Binary Latents,"We establish a theoretical link between evolutionary algorithms and variational parameter optimization of probabilistic generative models with binary hidden variables.While the novel approach is independent of the actual generative model, here we use two such models to investigate its applicability and scalability: a noisy-OR Bayes Net and Binary Sparse Coding.Learning of probabilistic generative models is first formulated as approximate maximum likelihood optimization using variational expectation maximization.We choose truncated posteriors as variational distributions in which discrete latent states serve as variational parameters.In the variational E-step,the latent states are thenoptimized according to a tractable free-energy objective. Given a data point, we can show that evolutionary algorithms can be used for the variational optimization loop by~considering the bit-vectors of the latent states as genomes of individuals, and by~defining the fitness of theindividuals as the joint probabilities given by the used generative model.As a proof of concept, we apply the novel evolutionary EM approach to the optimization of the parameters of noisy-OR Bayes nets and binary sparse coding on artificial and real data.Using point mutations and single-point cross-over for the evolutionary algorithm, we find that scalable variational EM algorithms are obtained which efficiently improve the data likelihood.In general we believe that, with the link established here, standard as well as recent results in the field of evolutionary optimization can be leveraged to address the difficult problem of parameter optimization in generative models.",We present Evolutionary EM as a novel algorithm for unsupervised training of generative models with binary latent variables that intimately connects variational EM with evolutionary optimizationThe paper presents a combination of evolutionary computation and variational EM for models with binary latent variables represented via a particle-based approximationThe paper makes an attempt to tightly integrate expectation-maximization training algorithms with evolutionary algorithms.
542,An Efficient Network for Predicting Time-Varying Distributions,"While deep neural networks have achieved groundbreaking prediction results in many tasks, there is a class of data where existing architectures are not optimal -- sequences of probability distributions.Performing forward prediction on sequences of distributions has many important applications.However, there are two main challenges in designing a network model for this task.First, neural networks are unable to encode distributions compactly as each node encodes just a real value.A recent work of Distribution Regression Network solved this problem with a novel network that encodes an entire distribution in a single node, resulting in improved accuracies while using much fewer parameters than neural networks.However, despite its compact distribution representation, DRN does not address the second challenge, which is the need to model time dependencies in a sequence of distributions.In this paper, we propose our Recurrent Distribution Regression Network which adopts a recurrent architecture for DRN.The combination of compact distribution representation and shared weights architecture across time steps makes RDRN suitable for modeling the time dependencies in a distribution sequence.Compared to neural networks and DRN, RDRN achieves the best prediction performance while keeping the network compact.",We propose an efficient recurrent network model for forward prediction on time-varying distributions.This paper proposes a method for creating neural nets that maps historical distributions onto distributions and applies the method to several distribution prediction tasks.Proposes a Reccurent Distribution Regression Network which uses a recurrent architecture upon a previous model Distribution Regression Network.This paper is on regressing over probability distributions by studying time varying distributions in a recurrent neural network setting
543,Graph Attention Networks,"We present graph attention networks, novel neural network architectures that operate on graph-structured data, leveraging masked self-attentional layers to address the shortcomings of prior methods based on graph convolutions or their approximations.By stacking layers in which nodes are able to attend over their neighborhoods' features, we enable specifying different weights to different nodes in a neighborhood, without requiring any kind of computationally intensive matrix operation or depending on knowing the graph structure upfront.In this way, we address several key challenges of spectral-based graph neural networks simultaneously, and make our model readily applicable to inductive as well as transductive problems.Our GAT models have achieved or matched state-of-the-art results across four established transductive and inductive graph benchmarks: the Cora, Citeseer and Pubmed citation network datasets, as well as a protein-protein interaction dataset.","A novel approach to processing graph-structured data by neural networks, leveraging attention over a node's neighborhood. Achieves state-of-the-art results on transductive citation network tasks and an inductive protein-protein interaction task."", 'This paper proposes a new method for classifying nodes of a graph, which can be used in semi-supervised scenarios and on a completely new graph. The paper introduces a neural network architecture to operate on graph-structured data named Graph Attention Networks.Provides a fair and almost comprehensive discussion of the state of art approaches to learning vector representations for the nodes of a graph."
544,N2N learning: Network to Network Compression via Policy Gradient Reinforcement Learning,"While bigger and deeper neural network architectures continue to advance the state-of-the-art for many computer vision tasks, real-world adoption of these networks is impeded by hardware and speed constraints.Conventional model compression methods attempt to address this problem by modifying the architecture manually or using pre-defined heuristics.Since the space of all reduced architectures is very large, modifying the architecture of a deep neural network in this way is a difficult task.In this paper, we tackle this issue by introducing a principled method for learning reduced network architectures in a data-driven way using reinforcement learning.Our approach takes a larger 'teacher' network as input and outputs a compressed 'student' network derived from the 'teacher' network."", ""In the first stage of our method, a recurrent policy network aggressively removes layers from the large 'teacher' model.In the second stage, another recurrent policy network carefully reduces the size of each remaining layer.The resulting network is then evaluated to obtain a reward -- a score based on the accuracy and compression of the network.Our approach uses this reward signal with policy gradients to train the policies to find a locally optimal student network.Our experiments show that we can achieve compression rates of more than 10x for models such as ResNet-34 while maintaining similar performance to the input 'teacher' network."", ""We also present a valuable transfer learning result which shows that policies which are pre-trained on smaller 'teacher' networks can be used to rapidly speed up training on larger 'teacher' networks.","A novel reinforcement learning based approach to compress deep neural networks with knowledge distillationThis paper proposes to use reinforcement learning instead of pre-defined heuristics to determine the structure of the compressed model in the knowledge distillation processIntroduces a principled way of network to network compression, which uses policy gradients for optimizing two policies which compress a strong teacher into a strong but smaller student model."
545,Harmonizing Maximum Likelihood with GANs for Multimodal Conditional Generation,"Recent advances in conditional image generation tasks, such as image-to-image translation and image inpainting, are largely accounted to the success of conditional GAN models, which are often optimized by the joint use of the GAN loss with the reconstruction loss.However, we reveal that this training recipe shared by almost all existing methods causes one critical side effect: lack of diversity in output samples.In order to accomplish both training stability and multimodal output generation, we propose novel training schemes with a new set of losses named moment reconstruction losses that simply replace the reconstruction loss.We show that our approach is applicable to any conditional generation tasks by performing thorough experiments on image-to-image translation, super-resolution and image inpainting using Cityscapes and CelebA dataset.Quantitative evaluations also confirm that our methods achieve a great diversity in outputs while retaining or even improving the visual fidelity of generated samples.","We prove that the mode collapse in conditional GANs is largely attributed to a mismatch between reconstruction loss and GAN loss and introduce a set of novel loss functions as alternatives for reconstruction loss.The paper proposes a modification to the traditional conditional GAN objective in order to promote diverse, multimodal generation of images. This paper proposes an alternative to L1/L2 errors that are used to augment adversarial losses when training conditional GANs."
546,On the difference between building and extracting patterns: a causal analysis of deep generative models.,"Generative models are important tools to capture and investigate the properties of complex empirical data.Recent developments such as Generative Adversarial Networks and Variational Auto-Encoders use two very similar, but , deep convolutional architectures, one to generate and one to extract information from data.Does learning the parameters of both architectures obey the same rules?We exploit the causality principle of independence of mechanisms to quantify how the weights of successive layers adapt to each other.Using the recently introduced Spectral Independence Criterion, we quantify the dependencies between the kernels of successive convolutional layers and show that those are more independent for the generative process than for information extraction, in line with results from the field of causal inference.In addition, our experiments on generation of human faces suggest that more independence between successive layers of generators results in improved performance of these architectures.","We use causal inference to characterise the architecture of generative modelsThis paper examines the nature of convolutional filters in the encoder and a decoder of a VAE, and a generator and a discriminator of a GAN.This work exploits the causality principle to quantify how the weights of successive layers adapt to each other."
547,Stochastic Gradient Descent Learns State Equations with Nonlinear Activations,"We study discrete time dynamical systems governed by the state equation.Here A,B are weight matrices, ϕ is an activation function, and is the input data.This relation is the backbone of recurrent neural networks which have broad applications in sequential learning tasks.We utilize stochastic gradient descent to learn the weight matrices from a finite input/state trajectory.We prove that SGD estimate linearly converges to the ground truth weights while using near-optimal sample size.Our results apply to increasing activations whose derivatives are bounded away from zero.The analysis is based oni) an SGD convergence result with nonlinear activations andii) careful statistical characterization of the state vector.Numerical experiments verify the fast convergence of SGD on ReLU and leaky ReLU in consistence with our theory.","We study the state equation of a recurrent neural network. We show that SGD can efficiently learn the unknown dynamics from few input/output observations under proper assumptions.The paper studies discrete-time dynamical systems with a non-linear state equation, proving that running SGD on a fixed-length trajectory gives logarithmic convergence.This work considers the problem of learning a non-linear dynamical system in which the output equals the state.This paper studies the ability of SGD to learn dynamics of a linear system and non-linear activation."
548,KNOWLEDGE DISTILL VIA LEARNING NEURON MANIFOLD,"Although deep neural networks show their extraordinary power in various tasks, they are not feasible for deploying such large models on embedded systems due to high computational cost and storage space limitation.The recent work knowledge distillation aims at transferring model knowledge from a well-trained teacher model to a small and fast student model which can significantly help extending the usage of large deep neural networks on portable platform.In this paper, we show that, by properly defining the neuron manifold of deep neuron network, we can significantly improve the performance of student DNN networks through approximating neuron manifold of powerful teacher network.To make this, we propose several novel methods for learning neuron manifold from DNN model.Empowered with neuron manifold knowledge, our experiments show the great improvement across a variety of DNN architectures and training data.Compared with other KD methods, our Neuron Manifold Transfer has best transfer ability of the learned features.",A new knowledge distill method for transfer learningThe work introduces a knowledge distillation method using the proposed neuron manifold concept. Proposes a knowledge distilling method in which neural manifold is taken as the transferred knowledge.
549,Slimmable Neural Networks,"We present a simple and general method to train a single neural network executable at different widths, permitting instant and adaptive accuracy-efficiency trade-offs at runtime.Instead of training individual networks with different width configurations, we train a shared network with switchable batch normalization.At runtime, the network can adjust its width on the fly according to on-device benchmarks and resource constraints, rather than downloading and offloading different models.Our trained networks, named slimmable neural networks, achieve similar ImageNet classification accuracy than individually trained models of MobileNet v1, MobileNet v2, ShuffleNet and ResNet-50 at different widths respectively.We also demonstrate better performance of slimmable models compared with individual ones across a wide range of applications including COCO bounding-box object detection, instance segmentation and person keypoint detection without tuning hyper-parameters.Lastly we visualize and discuss the learned features of slimmable networks.Code and models are available at: https://github.com/JiahuiYu/slimmable_networks","We present a simple and general method to train a single neural network executable at different widths (number of channels in a layer), permitting instant and adaptive accuracy-efficiency trade-offs at runtime.The paper proposes an idea of combining different size models together into one shared net, greatly improving performance for detectionThis paper trains a single network executable at different widths."
550,Learning Non-Metric Visual Similarity for Image Retrieval,"Measuring visualsimilarity between two or more instances within a data distribution is a fundamental task in many applications, specially in image retrieval.Theoretically, non-metric distances are able to generate a more complex and accurate similarity model than metric distances, provided that the non-linear data distribution is precisely captured by the similarity model.In this work, we analyze a simple approach for deep learning networks to be used as an approximation of non-metric similarity functions and we study how these models generalize across different image retrieval datasets.",Similarity network to learn a non-metric visual similarity estimation between a pair of imagesThe authors propose learning similarity measure for visual similarity and obtain by this an improvement in very well-known datasets of Oxford and Paris for image retrival.The paper argues that it is more suitable to use non-metric distances instead of metric distances.
551,Augmented Cyclic Adversarial Learning for Low Resource Domain Adaptation,"Training a model to perform a task typically requires a large amount of data from the domains in which the task will be applied.However, it is often the case that data are abundant in some domains but scarce in others.Domain adaptation deals with the challenge of adapting a model trained from a data-rich source domain to perform well in a data-poor target domain.In general, this requires learning plausible mappings between domains.CycleGAN is a powerful framework that efficiently learns to map inputs from one domain to another using adversarial training and a cycle-consistency constraint.However, the conventional approach of enforcing cycle-consistency via reconstruction may be overly restrictive in cases where one or more domains have limited training data.In this paper, we propose an augmented cyclic adversarial learning model that enforces the cycle-consistency constraint via an external task specific model, which encourages the preservation of task-relevant content as opposed to exact reconstruction.We explore digit classification in a low-resource setting in supervised, semi and unsupervised situation, as well as high resource unsupervised.In low-resource supervised setting, the results show that our approach improves absolute performance by 14% and 4% when adapting SVHN to MNIST and vice versa, respectively, which outperforms unsupervised domain adaptation methods that require high-resource unlabeled target domain. Moreover, using only few unsupervised target data, our approach can still outperforms many high-resource unsupervised models.Our model also outperforms on USPS to MNIST and synthetic digit to SVHN for high resource unsupervised adaptation.In speech domains, we similarly adopt a speech recognition model from each domain as the task specific model.Our approach improves absolute performance of speech recognition by 2% for female speakers in the TIMIT dataset, where the majority of training samples are from male voices.",A new cyclic adversarial learning augmented with auxiliary task model which improves domain adaptation performance in low resource supervised and unsupervised situations Proposes an extension of cycle-consistent adversatial adaptation methods in order to tackle domain adaptation where limited supervised target data is available.This paper introduces a domain adaptation approach based on the idea of Cyclic GAN and proposes two different algorithms.
552,Spectral Graph Wavelets for Structural Role Similarity in Networks,"Nodes residing in different parts of a graph can have similar structural roles within their local network topology.The identification of such roles provides key insight into the organization of networks and can also be used to inform machine learning on graphs.However, learning structural representations of nodes is a challenging unsupervised-learning task, which typically involves manually specifying and tailoring topological features for each node.Here we develop GraphWave, a method that represents each node’s local network neighborhood via a low-dimensional embedding by leveraging spectral graph wavelet diffusion patterns.We prove that nodes with similar local network neighborhoods will have similar GraphWave embeddings even though these nodes may reside in very different parts of the network.Our method scales linearly with the number of edges and does not require any hand-tailoring of topological features.We evaluate performance on both synthetic and real-world datasets, obtaining improvements of up to 71% over state-of-the-art baselines.","We develop a method for learning structural signatures in networks based on the diffusion of spectral graph wavelets. Using spectral graph wavelet diffusion patterns of a node's local meighbothood to embed the node in a low-dimensional space"", 'The paper derived a way to compare nodes in graph based on wavelet analysis of graph laplacian. "
553,UniNet: A Mixed Reality Driving Simulator,"Driving simulators play an important role in vehicle research.However, existing virtual reality simulators do not give users a true sense of presence.UniNet is our driving simulator, designed to allow users to interact with and visualize simulated traffic in mixed reality.It is powered by SUMO and Unity.UniNet's modular architecture allows us to investigate interdisciplinary research topics such as vehicular ad-hoc networks, human-computer interaction, and traffic management.We accomplish this by giving users the ability to observe and interact with simulated traffic in a high fidelity driving simulator.We present a user study that subjectively measures user's sense of presence in UniNet.Our findings suggest that our novel mixed reality system does increase this sensation.","A mixed reality driving simulator using stereo cameras and passthrough VR evaluated in a user study with 24 participants.Proposes a complicated system for driving simulation.This paper presents a mixed reality driving simulator setup to enhance the sensation of presenceProposes a mixed reality driving simulator that incorporates traffic generation and claims an enhanced ""presence"" due to an MR system."
554,Learning to Refer to 3D Objects with Natural Language,"Human world knowledge is both structured and flexible.When people see an object, they represent it not as a pixel array but as a meaningful arrangement of semantic parts.Moreover, when people refer to an object, they provide descriptions that are not merely true but also relevant in the current context.Here, we combine these two observations in order to learn fine-grained correspondences between language and contextually relevant geometric properties of 3D objects.To do this, we employed an interactive communication task with human participants to construct a large dataset containing natural utterances referring to 3D objects from ShapeNet in a wide variety of contexts.Using this dataset, we developed neural listener and speaker models with strong capacity for generalization.By performing targeted lesions of visual and linguistic input, we discovered that the neural listener depends heavily on part-related words and associates these words correctly with the corresponding geometric properties of objects, suggesting that it has learned task-relevant structure linking the two input modalities.We further show that a neural speaker that is `listener-aware' that plans its utterances according to how an imagined listener would interpret its words in context produces more discriminative referring expressions than an `listener-unaware' speaker, as measured by human performance in identifying the correct object.","How to build neural-speakers/listeners that learn fine-grained characteristics of 3D objects, from referential language.The authors provide a study on learning to refer to 3D objects, collecting a dataset of referential expressions and training several models by experimenting with a number of architectural choices"
555, Reasoning About Physical Interactions with Object-Oriented Prediction and Planning,"Object-based factorizations provide a useful level of abstraction for interacting with the world.Building explicit object representations, however, often requires supervisory signals that are difficult to obtain in practice.We present a paradigm for learning object-centric representations for physical scene understanding without direct supervision of object properties.Our model, Object-Oriented Prediction and Planning, jointly learns a perception function to map from image observations to object representations, a pairwise physics interaction function to predict the time evolution of a collection of objects, and a rendering function to map objects back to pixels.For evaluation, we consider not only the accuracy of the physical predictions of the model, but also its utility for downstream tasks that require an actionable representation of intuitive physics.After training our model on an image prediction task, we can use its learned representations to build block towers more complicated than those observed during training.","We present a framework for learning object-centric representations suitable for planning in tasks that require an understanding of physics.The paper presents a platform for predicting images of objects interacting with each other under the effect of gravitational forces. The paper presents a method that learns to reproduce 'block towers' from a given image."", 'Proposes a method which learns to reason on physical interaction of different objects with no supervison of object properties."
556,Global Optimality Conditions for Deep Neural Networks,"We study the error landscape of deep linear and nonlinear neural networks with the squared error loss.Minimizing the loss of a deep linear neural network is a nonconvex problem, and despite recent progress, our understanding of this loss surface is still incomplete.For deep linear networks, we present necessary and sufficient conditions for a critical point of the risk function to be a global minimum.Surprisingly, our conditions provide an efficiently checkable test for global optimality, while such tests are typically intractable in nonconvex optimization.We further extend these results to deep nonlinear neural networks and prove similar sufficient conditions for global optimality, albeit in a more limited function space setting.","We provide efficiently checkable necessary and sufficient conditions for global optimality in deep linear neural networks, with some initial extensions to nonlinear settings.The paper gives conditions for the global optimality of the loss function of deep linear neural networksThe paper gives theoretical results regarding the existence of local minima in the objective function of deep neural networks.Studies some theoretical properties of deep linear networks."
557,Recurrent Auto-Encoder Model for Multidimensional Time Series Representation,"Recurrent auto-encoder model can summarise sequential data through an encoder structure into a fixed-length vector and then reconstruct into its original sequential form through the decoder structure.The summarised information can be used to represent time series features.In this paper, we propose relaxing the dimensionality of the decoder output so that it performs partial reconstruction.The fixed-length vector can therefore represent features only in the selected dimensions.In addition, we propose using rolling fixed window approach to generate samples.The change of time series features over time can be summarised as a smooth trajectory path.The fixed-length vectors are further analysed through additional visualisation and unsupervised clustering techniques.This proposed method can be applied in large-scale industrial processes for sensors signal analysis purpose where clusters of the vector representations can be used to reflect the operating states of selected aspects of the industrial system.","Using recurrent auto-encoder model to extract multidimensional time series featuresThis writeup describes an application of recurrent autoencoder to analyze of multidimensional time seriesThe paper describes a sequence to sequence auto-encoder model which is used to learn sequence representations, showing that for their application, better performance is obtained when the network is only trained to reconstruct a subset of the data measurements. Proposes a strategy inspired by the recurrent auto-encoder model such that clustering multidimensional time series data can be performed based on context vectors."
558,Learning Multimodal Graph-to-Graph Translation for Molecule Optimization,"We view molecule optimization as a graph-to-graph translation problem.The goal is to learn to map from one molecular graph to another with better properties based on an available corpus of paired molecules.Since molecules can be optimized in different ways, there are multiple viable translations for each input graph.A key challenge is therefore to model diverse translation outputs.Our primary contributions include a junction tree encoder-decoder for learning diverse graph translations along with a novel adversarial training method for aligning distributions of molecules.Diverse output distributions in our model are explicitly realized by low-dimensional latent vectors that modulate the translation process.We evaluate our model on multiple molecule optimization tasks and show that our model outperforms previous state-of-the-art baselines by a significant margin.","We introduce a graph-to-graph encoder-decoder framework for learning diverse graph translations.Proposes a graph-to-graph translation model for molecule optimization inspired by matched molecular pair analysis.Extension of JT-VAE into the graph to graph translation scenario by adding the latent variable to capture multi-modality and an adversarial regularization in the latent spaceProposes a quite complex system, involving many different choices and components, for obtaining chemical compouds with improved properties starting from a given corpora."
559,Learning Neural PDE Solvers with Convergence Guarantees,"Partial differential equations are widely used across the physical and computational sciences.Decades of research and engineering went into designing fast iterative solution methods.Existing solvers are general purpose, but may be sub-optimal for specific classes of problems.In contrast to existing hand-crafted solutions, we propose an approach to learn a fast iterative solver tailored to a specific domain.We achieve this goal by learning to modify the updates of an existing solver using a deep neural network.Crucially, our approach is proven to preserve strong correctness and convergence guarantees.After training on a single geometry, our model generalizes to a wide variety of geometries and boundary conditions, and achieves 2-3 times speedup compared to state-of-the-art solvers.",We learn a fast neural solver for PDEs that has convergence guarantees.Develops a method to accelerate the finite difference method in solving PDEs and proposes a revised framework for fixed point iteration after discretization.The authors propose a linear method for speeding up PDE solvers.
560,FUNCTIONAL VARIATIONAL BAYESIAN NEURAL NETWORKS,"Variational Bayesian neural networks perform variational inference over weights, but it is difficult to specify meaningful priors and approximating posteriors in a high-dimensional weight space.We introduce functional variational Bayesian neural networks, which maximize an Evidence Lower BOund defined directly on stochastic processes, i.e. distributions over functions.We prove that the KL divergence between stochastic processes is equal to the supremum of marginal KL divergences over all finite sets of inputs.Based on this, we introduce a practical training objective which approximates the functional ELBO using finite measurement sets and the spectral Stein gradient estimator.With fBNNs, we can specify priors which entail rich structure, including Gaussian processes and implicit stochastic processes.Empirically, we find that fBNNs extrapolate well using various structured priors, provide reliable uncertainty estimates, and can scale to large datasets.",We perform functional variational inference on the stochastic processes defined by Bayesian neural networks.Fitting of variational Bayesian Neural Network approximations in functional form and considering matching to a stochastic process prior implicitly via samples.Presents a novel ELBO objective for training BNNs which allows for more meaningful priors to be encoded in the model rather than the less informative weight priors features in the literature.Presents a new variational inference algorithm for Bayesian neural network models where the prior is specified functionally rather than via a prior over weights. 
561,Poincare Glove: Hyperbolic Word Embeddings,"Words are not created equal.In fact, they form an aristocratic graph with a latent hierarchical structure that the next generation of unsupervised learned word embeddings should reveal.In this paper, justified by the notion of delta-hyperbolicity or tree-likeliness of a space, we propose to embed words in a Cartesian product of hyperbolic spaces which we theoretically connect to the Gaussian word embeddings and their Fisher geometry.This connection allows us to introduce a novel principled hypernymy score for word embeddings.Moreover, we adapt the well-known Glove algorithm to learn unsupervised word embeddings in this type of Riemannian manifolds.We further explain how to solve the analogy task using the Riemannian parallel transport that generalizes vector arithmetics to this new type of geometry.Empirically, based on extensive experiments, we prove that our embeddings, trained unsupervised, are the first to simultaneously outperform strong and popular baselines on the tasks of similarity, analogy and hypernymy detection.In particular, for word hypernymy, we obtain new state-of-the-art on fully unsupervised WBLESS classification accuracy.","We embed words in the hyperbolic space and make the connection with the Gaussian word embeddings.This paper adapts the Glove word embedding to a hyperbolic space given by the Poincare half-plane modelThis paper proposes an approach to implement a GLOVE-based hyperbolic word embedding model, which is optimized via the Riemannian Optimization methods."
562,How to learn (and how not to learn) multi-hop reasoning with memory networks,"Answering questions about a text frequently requires aggregating information from multiple places in that text.End-to-end neural network models, the dominant approach in the current literature, can theoretically learn how to distill and manipulate representations of the text without explicit supervision about how to do so.We investigate a canonical architecture for this task, the memory network, and analyze how effective it really is in the context of three multi-hop reasoning settings.In a simple synthetic setting, the path-finding task of the bAbI dataset, the model fails to learn the correct reasoning without additional supervision of its attention mechanism.However, with this supervision, it can perform well.On a real text dataset, WikiHop, the memory network gives nearly state-of-the-art performance, but does so without using its multi-hop capabilities.A tougher anonymized version of the WikiHop dataset is qualitatively similar to bAbI: the model fails to perform well unless it has additional supervision.We hypothesize that many ""multi-hop"" architectures do not truly learn this reasoning as advertised, though they could learn this reasoning if appropriately supervised.","Memory Networks do not learn multi-hop reasoning unless we supervise them. Claims multi-hop reasoning is not easy to learn directly and requires direct supervision and doing well on WikiHop doesn't necessarily mean the model is actually learning to hop."", 'The paper proposes to investigate the well-known problem of memory network learning and more precisely the difficulty of the attention learning supervision with such models.This paper argues that memory network fails to learn reasonable multi-hop reasoning."
563,Generative networks as inverse problems with Scattering transforms,"Generative Adversarial Nets and Variational Auto-Encoders provide impressive image generations from Gaussian white noise, but the underlying mathematics are not well understood.We compute deep convolutional network generators by inverting a fixed embedding operator.Therefore, they do not require to be optimized with a discriminator or an encoder.The embedding is Lipschitz continuous to deformations so that generators transform linear interpolations between input white noise vectors into deformations between output images.This embedding is computed with a wavelet Scattering transform.Numerical experiments demonstrate that the resulting Scattering generators have similar properties as GANs or VAEs, without learning a discriminative network or an encoder.",We introduce generative networks that do not require to be learned with a discriminator or an encoder; they are obtained by inverting a special embedding operator defined by a wavelet Scattering transform.Introduces scattering transforms as image generative models in the context of Generative Adversarial Networks and suggest why they could be seen as Gaussianization transforms with controlled information loss and invertibility. The paper proposes a generative model for images that does no require to learn a discriminator (as in GAN’s) or learned embedding.
564,Neural Speed Reading with Structural-Jump-LSTM,"Recurrent neural networks can model natural language by sequentially ''reading'' input tokens and outputting a distributed representation of each token.Due to the sequential nature of RNNs, inference time is linearly dependent on the input length, and all inputs are read regardless of their importance.Efforts to speed up this inference, known as ''neural speed reading'', either ignore or skim over part of the input.We present Structural-Jump-LSTM: the first neural speed reading model to both skip and jump text during inference.The model consists of a standard LSTM and two agents: one capable of skipping single words when reading, and one capable of exploiting punctuation structure, sentence end symbols, or end of text markers) to jump ahead after reading a word.A comprehensive experimental evaluation of our model against all five state-of-the-art neural reading models shows thatStructural-Jump-LSTM achieves the best overall floating point operations reduction, while keeping the same accuracy or even improving it compared to a vanilla LSTM that reads the whole text.","We propose a new model for neural speed reading that utilizes the inherent punctuation structure of a text to define effective jumping and skipping behavior.The paper proposes a Structural-Jump-LSTM model to speed up machine reading with two agents instead of oneProposes a novel model for neural speed reading in which the new reader has the ability to skip a word or sequence of words.The paper proposes a fast-reading method using skip and jump actions, showing that the proposed method is as accurate as LSTM but uses much less computation."
565,Purchase as Reward : Session-based  Recommendation by Imagination Reconstruction,"One of the key challenges of session-based recommender systems is to enhance users’ purchase intentions.In this paper, we formulate the sequential interactions between user sessions and a recommender agent as a Markov Decision Process.In practice, the purchase reward is delayed and sparse, and may be buried by clicks, making it an impoverished signal for policy learning.Inspired by the prediction error minimization and embodied cognition, we propose a simple architecture to augment reward, namely Imagination Reconstruction Network.Specifically, IRN enables the agent to explore its environment and learn predictive representations via three key components.The imagination core generates predicted trajectories, i.e., imagined items that users may purchase.The trajectory manager controls the granularity of imagined trajectories using the planning strategies, which balances the long-term rewards and short-term rewards.To optimize the action policy, the imagination-augmented executor minimizes the intrinsic imagination error of simulated trajectories by self-supervised reconstruction, while maximizing the extrinsic reward using model-free algorithms.Empirically, IRN promotes quicker adaptation to user interest, and shows improved robustness to the cold-start scenario and ultimately higher purchase performance compared to several baselines.Somewhat surprisingly, IRN using only the purchase reward achieves excellent next-click prediction performance, demonstrating that the agent can ""guess what you like"" via internal planning.",We propose the IRN architecture to augment sparse and delayed purchase reward for session-based recommendation.The paper proposes improving the performance of recommendation systems through reinforcement learning by using an Imagination Reconstruction Network.The paper presents a session-based recommendation approach by focusing on user purchases instead of clicks. 
566,Ensemble Robustness and Generalization of Stochastic Deep Learning Algorithms,"The question why deep learning algorithms generalize so well has attracted increasingresearch interest.However, most of the well-established approaches,such as hypothesis capacity, stability or sparseness, have not provided completeexplanations.In this work, we focuson the robustness approach, i.e., if the error of a hypothesiswill not change much due to perturbations of its training examples, then itwill also generalize well.As most deep learning algorithms are stochastic, we revisit the robustnessarguments of Xu & Mannor, and introduce a new approach – ensemblerobustness – that concerns the robustness of a population of hypotheses.Throughthe lens of ensemble robustness, we reveal that a stochastic learning algorithm cangeneralize well as long as its sensitiveness to adversarial perturbations is boundedin average over training examples.Moreover, an algorithm may be sensitive tosome adversarial examples but still generalize well.Tosupport our claims, we provide extensive simulations for different deep learningalgorithms and different network architectures exhibiting a strong correlation betweenensemble robustness and the ability to generalize.","Explaining the generalization of stochastic deep learning algorithms, theoretically and empirically, via ensemble robustness This paper presents an adaptation of the algorithmic robustness of Xu&Mannor'12 and presents learning bounds and an experimental showing correlation between empirical ensemble robustness and generalization error. "", 'Proposes a study of the generalization ability of deep learning algorithms using an extension of notion of stability called ensemble robustness and gives bounds on generalization error of a randomized algorithm in terms of stability parameter and provides empirical study attempting to connect theory with practice.The paper studied the generalization ability of learning algorithms from the robustness viewpoint in a deep learning context"
567,Few-shot Autoregressive Density Estimation: Towards Learning to Learn Distributions,"Deep autoregressive models have shown state-of-the-art performance in density estimation for natural images on large-scale datasets such as ImageNet. However, such models require many thousands of gradient-based weight updates and unique image examples for training.Ideally, the models would rapidly learn visual concepts from only a handful of examples, similar to the manner in which humans learns across many vision tasks. In this paper, we show how 1) neural attention and 2) meta learning techniques can be used in combination with autoregressive models to enable effective few-shot density estimation.Our proposed modifications to PixelCNN result in state-of-the art few-shot density estimation on the Omniglot dataset. Furthermore, we visualize the learned attention policy and find that it learns intuitive algorithms for simple tasks such as image mirroring on ImageNet and handwriting on Omniglot without supervision.Finally, we extend the model to natural images and demonstrate few-shot image generation on the Stanford Online Products dataset.","Few-shot learning PixelCNNThe paper proposes on using density estimation when the availability of training data is low by using a meta-learning model.This paper considers the problem of one/few-shot density estimation, using metalearning techniques that have been applied to one/few-shot supervised learningThe paper focuses on few shot learning with autoregressive density estimation and improves PixelCNN with neural attention and meta learning techniques."
568,SGD Learns Over-parameterized Networks that Provably Generalize on Linearly Separable Data,"Neural networks exhibit good generalization behavior in theover-parameterized regime, where the number of network parametersexceeds the number of observations.Nonetheless,current generalization bounds for neural networks fail to explain thisphenomenon.In an attempt to bridge this gap, we study the problem oflearning a two-layer over-parameterized neural network, when the data is generated by a linearly separable function.In the case where the network has LeakyReLU activations, we provide both optimization and generalization guarantees for over-parameterized networks.Specifically, we prove convergence rates of SGD to a globalminimum and provide generalization guarantees for this global minimumthat are independent of the network size.Therefore, our result clearly shows that the use of SGD for optimization both finds a global minimum, and avoids overfitting despite the high capacity of the model.This is the first theoretical demonstration that SGD can avoid overfitting, when learning over-specified neural network classifiers.","We show that SGD learns two-layer over-parameterized neural networks with Leaky ReLU activations that provably generalize on linearly separable data.The paper studies overparameterised models being able to learn well-generalising solutions by using a 1-hidden layer network with fixed output layer.This paper shows that on linearly seperabel data, SGD on an overparameterized network can still lean a classifier that provably generalizes."
569,InfoBot: Transfer and Exploration via the Information Bottleneck,"A central challenge in reinforcement learning is discovering effective policies for tasks where rewards are sparsely distributed.We postulate that in the absence of useful reward signals, an effective exploration strategy should seek out.These states lie at critical junctions in the state space from where the agent can transition to new, potentially unexplored regions.We propose to learn about decision states from prior experience.By training a goal-conditioned model with an information bottleneck, we can identify decision states by examining where the model accesses the goal state through the bottleneck.We find that this simple mechanism effectively identifies decision states, even in partially observed settings.In effect, the model learns the sensory cues that correlate with potential subgoals.In new environments, this model can then identify novel subgoals for further exploration, guiding the agent through a sequence of potential decision states and through new regions of the state space.",Training agents with goal-policy information bottlenecks promotes transfer and yields a powerful exploration bonusProposes regularizing standard RL losses with the negative conditional mutual information for policy search in a multi-goal RL setting.This paper proposes the concept of decision state and proposes a KL divergence regularization to learn the structure of the tasks to use this information to encourage the policy to visit the decision states.The paper proposes a method of regularising goal-conditioned policies with a mutual information term. 
570,Guided Evolutionary Strategies: Escaping the curse of dimensionality in random search,"Many applications in machine learning require optimizing a function whose true gradient is unknown, but where surrogate gradient information is available instead.This arises when an approximate gradient is easier to compute than the full gradient, or when a true gradient is intractable and is replaced with a surrogate.We propose Guided Evolutionary Strategies, a method for optimally using surrogate gradient directions along with random search.We define a search distribution for evolutionary strategies that is elongated along a subspace spanned by the surrogate gradients.This allows us to estimate a descent direction which can then be passed to a first-order optimizer.We analytically and numerically characterize the tradeoffs that result from tuning how strongly the search distribution is stretched along the guiding subspace, and use this to derive a setting of the hyperparameters that works well across problems.Finally, we apply our method to example problems including truncated unrolled optimization and training neural networks with discrete variables, demonstrating improvement over both standard evolutionary strategies and first-order methods.We provide a demo of Guided ES at: redacted URL","We propose an optimization method for when only biased gradients are available--we define a new gradient estimator for this scenario, derive the bias and variance of this estimator, and apply it to example problems.The authors propose an approach that combines random search with the surrogate gradient information and give a discussion on variance-bias trade-off as well as a discussion on hyperparameter optimization. The paper proposes a method to improve random search by building a subspace of the previous k surrogate gradients.This paper attempts accelerating the OpenAI type evolution by introducing a non-isotrophic distribution with a covariance matrix in the form I + UU^t and external information such as a surrogate gradient to determine U"
571,Learning Localized Generative Models for 3D Point Clouds via Graph Convolution,"Point clouds are an important type of geometric data and have widespread use in computer graphics and vision.However, learning representations for point clouds is particularly challenging due to their nature as being an unordered collection of points irregularly distributed in 3D space.Graph convolution, a generalization of the convolution operation for data defined over graphs, has been recently shown to be very successful at extracting localized features from point clouds in supervised or semi-supervised tasks such as classification or segmentation.This paper studies the unsupervised problem of a generative model exploiting graph convolution.We focus on the generator of a GAN and define methods for graph convolution when the graph is not known in advance as it is the very output of the generator.The proposed architecture learns to generate localized features that approximate graph embeddings of the output geometry.We also study the problem of defining an upsampling layer in the graph-convolutional generator, such that it learns to exploit a self-similarity prior on the data distribution to sample more effectively.",A GAN using graph convolution operations with dynamically computed graphs from hidden featuresThe paper proposes a version of GANs specifically designed for generating point clouds with the core contribution of the work the upsampling operation.This paper proposes graph-convolutional GANs for irregular 3D point clouds that learn domain and features at the same time.
572,Better Generalization with On-the-fly Dataset Denoising,"Memorization in over-parameterized neural networks can severely hurt generalization in the presence of mislabeled examples.However, mislabeled examples are to hard avoid in extremely large datasets.We address this problem using the implicit regularization effect of stochastic gradient descent with large learning rates, which we find to be able to separate clean and mislabeled examples with remarkable success using loss statistics.We leverage this to identify and on-the-fly discard mislabeled examples using a threshold on their losses.This leads to On-the-fly Data Denoising, a simple yet effective algorithm that is robust to mislabeled examples, while introducing almost zero computational overhead.Empirical results demonstrate the effectiveness of ODD on several datasets containing artificial and real-world mislabeled examples.",We introduce a fast and easy-to-implement algorithm that is robust to dataset noise.The paper aims to remove potential examples with label noise by discarding the ones with large losses in the training procedure.
573,Combinatorial Attacks on Binarized Neural Networks,"Binarized Neural Networks have recently attracted significant interest due to their computational efficiency.Concurrently, it has been shown that neural networks may be overly sensitive to attacks"" -- tiny adversarial changes in the input -- which may be detrimental to their use in safety-critical domains.Designing attack algorithms that effectively fool trained models is a key step towards learning robust neural networks.The discrete, non-differentiable nature of BNNs, which distinguishes them from their full-precision counterparts, poses a challenge to gradient-based attacks.In this work, we study the problem of attacking a BNN through the lens of combinatorial and integer optimization.We propose a Mixed Integer Linear Programming formulation of the problem.While exact and flexible, the MILP quickly becomes intractable as the network and perturbation space grow.To address this issue, we propose IProp, a decomposition-based algorithm that solves a sequence of much smaller MILP problems.Experimentally, we evaluate both proposed methods against the standard gradient-based attack on MNIST and Fashion-MNIST, and show that IProp performs favorably compared to PGD, while scaling beyond the limits of the MILP.",Gradient-based attacks on binarized neural networks are not effective due to the non-differentiability of such networks; Our IPROP algorithm solves this problem using integer optimizationProposes a new target propagation style algorithm to generate strong adversarial attacks on binarized neural networks.This paper proposed a new attack algorithm based on MILP on binary neural networks.This paper presents an algorithm to find adversarial attacks to binary neural networks which iteratively finds desired representations layer by layer from the top to the input and is more efficient than solving the full mixed integer linear programming (MILP) solver.
574,Improved Language Modeling by Decoding the Past,"Highly regularized LSTMs achieve impressive results on several benchmark datasets in language modeling.We propose a new regularization method based on decoding the last token in the context using the predicted distribution of the next token.This biases the model towards retaining more contextual information, in turn improving its ability to predict the next token.With negligible overhead in the number of parameters and training time, our Past Decode Regularization method achieves a word level perplexity of 55.6 on the Penn Treebank and 63.5 on the WikiText-2 datasets using a single softmax.We also show gains by using PDR in combination with a mixture-of-softmaxes, achieving a word level perplexity of 53.8 and 60.5 on these datasets.In addition, our method achieves 1.169 bits-per-character on the Penn Treebank Character dataset for character level language modeling.These results constitute a new state-of-the-art in their respective settings.",Decoding the last token in the context using the predicted next token distribution acts as a regularizer and improves language modeling.The authors introduce the idea of past decoding for the purpose of regularization for improved perplexity on Penn TreebankProposes an additional loss term to use when training an LSTM LM and shows that by adding this loss term they can achieve SOTA perplexity on a number of LM benchmarks.Suggests a new regularization technique which can be added on top of those used in AWD-LSTM of Merity et al. (2017) with little overhead.
575,Discovering Order in Unordered Datasets: Generative Markov Networks,"The assumption that data samples are independently identically distributed is the backbone of many learning algorithms.Nevertheless, datasets often exhibit rich structures in practice, and we argue that there exist some unknown orders within the data instances.Aiming to find such orders, we introduce a novel Generative Markov Network which we use to extract the order of data instances automatically.Specifically, we assume that the instances are sampled from a Markov chain.Our goal is to learn the transitional operator of the chain as well as the generation order by maximizing the generation probability under all possible data permutations.One of our key ideas is to use neural networks as a soft lookup table for approximating the possibly huge, but discrete transition matrix.This strategy allows us to amortize the space complexity with a single model and make the transitional operator generalizable to unseen instances.To ensure the learned Markov chain is ergodic, we propose a greedy batch-wise permutation scheme that allows fast training. Empirically, we evaluate the learned Markov chain by showing that GMNs are able to discover orders among data instances and also perform comparably well to state-of-the-art methods on the one-shot recognition benchmark task.",Propose to observe implicit orders in datasets in a generative model viewpoint.The authors deal with the problem of implicit ordering in a dataset and the challenge of recovering it and propose to learn a distance-metric-free model that assumes a Markov chain as the generative mechanism of the data The paper proposes “Generative Markov Networks” - a deep-learning-based approach to modeling sequences and discovering order in datasets.Proposes learning the order of an unordered data sample by learning a Markov chain.
576,On the Discrimination-Generalization Tradeoff in GANs,"Generative adversarial training can be generally understood as minimizing certain moment matching loss defined by a set of discriminator functions, typically neural networks.The discriminator set should be large enough to be able to uniquely identify the true distribution, and also be small enough to go beyond memorizing samples.In this paper, we show that a discriminator set is guaranteed to be discriminative whenever its linear span is dense in the set of bounded continuous functions.This is a very mild condition satisfied even by neural networks with a single neuron.Further, we develop generalization bounds between the learned distribution and true distribution under different evaluation metrics.When evaluated with neural distance, our bounds show that generalization is guaranteed as long as the discriminator set is small enough, regardless of the size of the generator or hypothesis set.When evaluated with KL divergence, our bound provides an explanation on the counter-intuitive behaviors of testing likelihood in GAN training.Our analysis sheds lights on understanding the practical performance of GANs.",This paper studies the discrimination and generalization properties of GANs when the discriminator set is a restricted function class like neural networks.Balances capacities of generator and discriminator classes in GANs by guaranteeing that induced IPMs are metrics and not pseudo metricsThis paper provides a mathematical analysis of the role of the size of the adversary/discriminator set in GANs
577,Fixup Initialization: Residual Learning Without Normalization,"Normalization layers are a staple in state-of-the-art deep neural network architectures.They are widely believed to stabilize training, enable higher learning rate, accelerate convergence and improve generalization, though the reason for their effectiveness is still an active research topic.In this work, we challenge the commonly-held beliefs by showing that none of the perceived benefits is unique to normalization.Specifically, we propose fixed-update initialization, an initialization motivated by solving the exploding and vanishing gradient problem at the beginning of training via properly rescaling a standard initialization.We find training residual networks with Fixup to be as stable as training with normalization -- even for networks with 10,000 layers.Furthermore, with proper regularization, Fixup enables residual networks without normalization to achieve state-of-the-art performance in image classification and machine translation.","All you need to train deep residual networks is a good initialization; normalization layers are not necessary.A method is presented for initialization and normalization of deep residual networks. This is based on observations of forward and backward explosion in such networks. The method performance is on par with the best results obtained by other networks with more explicit normalization.The authors propose a novel way to initialize residual networks, which is motivated by the need to avoid exploding/vanishing gradients.Proposes a new initialization method used to train very deep RedNets without using batch-norm."
578,Toward learning better metrics for sequence generation training with policy gradient,"Designing a metric manually for unsupervised sequence generation tasks, such as text generation, is essentially difficult.In a such situation, learning a metric of a sequence from data is one possible solution.The previous study, SeqGAN, proposed the framework for unsupervised sequence generation, in which a metric is learned from data, and a generator is optimized with regard to the learned metric with policy gradient, inspired by generative adversarial nets and reinforcement learning.In this paper, we make two proposals to learn better metric than SeqGAN's: partial reward function and expert-based reward function training.The partial reward function is a reward function for a partial sequence of a certain length.SeqGAN employs a reward function for completed sequence only.By combining long-scale and short-scale partial reward functions, we expect a learned metric to be able to evaluate a partial correctness as well as a coherence of a sequence, as a whole.In expert-based reward function training, a reward function is trained to discriminate between an expert sequence and a fake sequence that is produced by editing an expert sequence.Expert-based reward function training is not a kind of GAN frameworks.This makes the optimization of the generator easier.We examine the effect of the partial reward function and expert-based reward function training on synthetic data and real text data, and show improvements over SeqGAN and the model trained with MLE.Specifically, whereas SeqGAN gains 0.42 improvement of NLL over MLE on synthetic data, our best model gains 3.02 improvement, and whereas SeqGAN gains 0.029 improvement of BLEU over MLE, our best model gains 0.250 improvement.","This paper aims to learn a better metric for unsupervised learning, such as text generation, and shows a significant improvement over SeqGAN.Describes an approach to generating time sequences by learning state-action values, where the state is the sequence generated so far, and the action is the choice of the next value. This paper considers the problem of improving sequence generation by learning better metrics, specifically the exposure bias problem"
579,Learning Generative Models with Locally Disentangled Latent Factors,"One of the most successful techniques in generative models has been decomposing a complicated generation task into a series of simpler generation tasks. For example, generating an image at a low resolution and then learning to refine that into a high resolution image often improves results substantially. Here we explore a novel strategy for decomposing generation for complicated objects in which we first generate latent variables which describe a subset of the observed variables, and then map from these latent variables to the observed space. We show that this allows us to achieve decoupled training of complicated generative models and present both theoretical and experimental results supporting the benefit of such an approach. ","Decompose the task of learning a generative model into learning disentangled latent factors for subsets of the data and then learning the joint over those latent factors. Locally Disentangled Factors for hierarchical latent variable generative model, which can be seen as a hierarchical variant of Adversarially Learned InferenceThe paper investigates the potential of hierarchical latent variable models for generating images and image sequences and proposes to train several ALI models stacked on top of each other to create a hierarchical representation of the data.The paper aims to learn the hierarchies for training GAN in a hierarchical optimization schedule directly instead of being designed by a human"
580,Learning Grounded Sentence Representations by Jointly Using Video and Text Information,"Visual grounding of language is an active research field aiming at enriching text-based representations with visual information.In this paper, we propose a new way to leverage visual knowledge for sentence representations.Our approach transfers the structure of a visual representation space to the textual space by using two complementary sources of information: the cluster information: the implicit knowledge that two sentences associated with the same visual content describe the same underlying reality and the perceptual information contained within the structure of the visual space.We use a joint approach to encourage beneficial interactions during training between textual, perceptual, and cluster information.We demonstrate the quality of the learned representations on semantic relatedness, classification, and cross-modal retrieval tasks.","We propose a joint model to incorporate visual knowledge in sentence representationsThe paper proposes a method to use videos paired with captions to improve sentence embeddingsThis submission proposes a model for sentence learning sentence representations that are grounded, based on associated video data.Proposes a method for improving text-based sentence embeddings through a joint multimodal framework."