Datasets:

kardosdrur

/

arxiv_multiclass

like 0

We present an unsupervised learning framework for decomposing images into layers of automatically discovered object models. Contrary to recent approaches that model image layers with autoencoder networks, we represent them as explicit transformations of a small set of prototypical images. Our model has three main components: (i) a set of object prototypes in the form of learnable images with a transparency channel, which we refer to as sprites; (ii) differentiable parametric functions predicting occlusions and transformation parameters necessary to instantiate the sprites in a given image; (iii) a layered image formation model with occlusion for compositing these instances into complete images including background. By jointly learning the sprites and occlusion/transformation predictors to reconstruct images, our approach not only yields accurate layered image decompositions, but also identifies object categories and instance parameters. We first validate our approach by providing results on par with the state of the art on standard multi-object synthetic benchmarks (Tetrominoes, Multi-dSprites, CLEVR6). We then demonstrate the applicability of our model to real images in tasks that include clustering (SVHN, GTSRB), cosegmentation (Weizmann Horse) and object discovery from unfiltered social network images. To the best of our knowledge, our approach is the first layered image decomposition algorithm that learns an explicit and shared concept of object type, and is robust enough to be applied to real images.

We introduce a deep and light-weight transformer, DeLighT, that delivers similar or better performance than standard transformer-based models with significantly fewer parameters. DeLighT more efficiently allocates parameters both (1) within each Transformer block using the DeLighT transformation, a deep and light-weight transformation, and (2) across blocks using block-wise scaling, which allows for shallower and narrower DeLighT blocks near the input and wider and deeper DeLighT blocks near the output. Overall, DeLighT networks are 2.5 to 4 times deeper than standard transformer models and yet have fewer parameters and operations. Experiments on benchmark machine translation and language modeling tasks show that DeLighT matches or improves the performance of baseline Transformers with 2 to 3 times fewer parameters on average. Our source code is available at: \url{https://github.com/sacmehta/delight}

Removing or repairing the imperfections of a digital images or videos is a very active and attractive field of research belonging to the image inpainting technique. This later has a wide range of applications, such as removing scratches in old photographic image, removing text and logos or creating cartoon and artistic effects. In this paper, we propose an efficient method to repair a damaged image based on a non linear diffusion tensor. The idea is to track perfectly the local geometry of the damaged image and allowing diffusion only in the isophotes curves direction. To illustrate the effective performance of our method, we present some experimental results on test and real photographic color images

Generative models often use human evaluations to measure the perceived quality of their outputs. Automated metrics are noisy indirect proxies, because they rely on heuristics or pretrained embeddings. However, up until now, direct human evaluation strategies have been ad-hoc, neither standardized nor validated. Our work establishes a gold standard human benchmark for generative realism. We construct Human eYe Perceptual Evaluation (HYPE) a human benchmark that is (1) grounded in psychophysics research in perception, (2) reliable across different sets of randomly sampled outputs from a model, (3) able to produce separable model performances, and (4) efficient in cost and time. We introduce two variants: one that measures visual perception under adaptive time constraints to determine the threshold at which a model's outputs appear real (e.g. 250ms), and the other a less expensive variant that measures human error rate on fake and real images sans time constraints. We test HYPE across six state-of-the-art generative adversarial networks and two sampling techniques on conditional and unconditional image generation using four datasets: CelebA, FFHQ, CIFAR-10, and ImageNet. We find that HYPE can track model improvements across training epochs, and we confirm via bootstrap sampling that HYPE rankings are consistent and replicable.

A central mechanism in machine learning is to identify, store, and recognize patterns. How to learn, access, and retrieve such patterns is crucial in Hopfield networks and the more recent transformer architectures. We show that the attention mechanism of transformer architectures is actually the update rule of modern Hopfield networks that can store exponentially many patterns. We exploit this high storage capacity of modern Hopfield networks to solve a challenging multiple instance learning (MIL) problem in computational biology: immune repertoire classification. Accurate and interpretable machine learning methods solving this problem could pave the way towards new vaccines and therapies, which is currently a very relevant research topic intensified by the COVID-19 crisis. Immune repertoire classification based on the vast number of immunosequences of an individual is a MIL problem with an unprecedentedly massive number of instances, two orders of magnitude larger than currently considered problems, and with an extremely low witness rate. In this work, we present our novel method DeepRC that integrates transformer-like attention, or equivalently modern Hopfield networks, into deep learning architectures for massive MIL such as immune repertoire classification. We demonstrate that DeepRC outperforms all other methods with respect to predictive performance on large-scale experiments, including simulated and real-world virus infection data, and enables the extraction of sequence motifs that are connected to a given disease class. Source code and datasets: https://github.com/ml-jku/DeepRC

We propose a new approach to interactive image segmentation based on some properties of a family of quadratic optimization problems related to dominant sets, a well-known graph-theoretic notion of a cluster which generalizes the concept of a maximal clique to edge-weighted graphs. In particular, we show that by properly controlling a regularization parameter which determines the structure and the scale of the underlying problem, we are in a position to extract groups of dominant-set clusters which are constrained to contain user-selected elements. The resulting algorithm can deal naturally with any type of input modality, including scribbles, sloppy contours, and bounding boxes, and is able to robustly handle noisy annotations on the part of the user. Experiments on standard benchmark datasets show the effectiveness of our approach as compared to state-of-the-art algorithms on a variety of natural images under several input conditions.

Detecting partially occluded objects is a difficult task. Our experimental results show that deep learning approaches, such as Faster R-CNN, are not robust at object detection under occlusion. Compositional convolutional neural networks (CompositionalNets) have been shown to be robust at classifying occluded objects by explicitly representing the object as a composition of parts. In this work, we propose to overcome two limitations of CompositionalNets which will enable them to detect partially occluded objects: 1) CompositionalNets, as well as other DCNN architectures, do not explicitly separate the representation of the context from the object itself. Under strong object occlusion, the influence of the context is amplified which can have severe negative effects for detection at test time. In order to overcome this, we propose to segment the context during training via bounding box annotations. We then use the segmentation to learn a context-aware CompositionalNet that disentangles the representation of the context and the object. 2) We extend the part-based voting scheme in CompositionalNets to vote for the corners of the object's bounding box, which enables the model to reliably estimate bounding boxes for partially occluded objects. Our extensive experiments show that our proposed model can detect objects robustly, increasing the detection performance of strongly occluded vehicles from PASCAL3D+ and MS-COCO by 41% and 35% respectively in absolute performance relative to Faster R-CNN.

Since many real-world data can be described from multiple views, multi-view learning has attracted considerable attention. Various methods have been proposed and successfully applied to multi-view learning, typically based on matrix factorization models. Recently, it is extended to the deep structure to exploit the hierarchical information of multi-view data, but the view-specific features and the label information are seldom considered. To address these concerns, we present a partially shared semi-supervised deep matrix factorization model (PSDMF). By integrating the partially shared deep decomposition structure, graph regularization and the semi-supervised regression model, PSDMF can learn a compact and discriminative representation through eliminating the effects of uncorrelated information. In addition, we develop an efficient iterative updating algorithm for PSDMF. Extensive experiments on five benchmark datasets demonstrate that PSDMF can achieve better performance than the state-of-the-art multi-view learning approaches. The MATLAB source code is available at https://github.com/libertyhhn/PartiallySharedDMF.

In recent years, the idea of using morphological operations as networks has received much attention. Mathematical morphology provides very efficient and useful image processing and image analysis tools based on basic operators like dilation and erosion, defined in terms of kernels. Many other morphological operations are built up using the dilation and erosion operations. Although the learning of structuring elements such as dilation or erosion using the backpropagation algorithm is not new, the order and the way these morphological operations are used is not standard. In this paper, we have theoretically analyzed the use of morphological operations for processing 1D feature vectors and shown that this gets extended to the 2D case in a simple manner. Our theoretical results show that a morphological block represents a sum of hinge functions. Hinge functions are used in many places for classification and regression tasks (Breiman (1993)). We have also proved a universal approximation theorem -- a stack of two morphological blocks can approximate any continuous function over arbitrary compact sets. To experimentally validate the efficacy of this network in real-life applications, we have evaluated its performance on satellite image classification datasets since morphological operations are very sensitive to geometrical shapes and structures. We have also shown results on a few tasks like segmentation of blood vessels from fundus images, segmentation of lungs from chest x-ray and image dehazing. The results are encouraging and further establishes the potential of morphological networks.

In this paper, we focus on obtaining 2D and 3D labels, as well as track IDs for objects on the road with the help of a novel 3D Bounding Box Annotation Toolbox (3D BAT). Our open source, web-based 3D BAT incorporates several smart features to improve usability and efficiency. For instance, this annotation toolbox supports semi-automatic labeling of tracks using interpolation, which is vital for downstream tasks like tracking, motion planning and motion prediction. Moreover, annotations for all camera images are automatically obtained by projecting annotations from 3D space into the image domain. In addition to the raw image and point cloud feeds, a Masterview consisting of the top view (bird's-eye-view), side view and front views is made available to observe objects of interest from different perspectives. Comparisons of our method with other publicly available annotation tools reveal that 3D annotations can be obtained faster and more efficiently by using our toolbox.

Computer-aided diagnosis (CADx) algorithms in medicine provide patient-specific decision support for physicians. These algorithms are usually applied after full acquisition of high-dimensional multimodal examination data, and often assume feature-completeness. This, however, is rarely the case due to examination costs, invasiveness, or a lack of indication. A sub-problem in CADx, which to our knowledge has received very little attention among the CADx community so far, is to guide the physician during the entire peri-diagnostic workflow, including the acquisition stage. We model the following question, asked from a physician's perspective: "Given the evidence collected so far, which examination should I perform next, in order to achieve the most accurate and efficient diagnostic prediction?". In this work, we propose a novel approach which is enticingly simple: use dropout at the input layer, and integrated gradients of the trained network at test-time to attribute feature importance dynamically. We validate and explain the effectiveness of our proposed approach using two public medical and two synthetic datasets. Results show that our proposed approach is more cost- and feature-efficient than prior approaches and achieves a higher overall accuracy. This directly translates to less unnecessary examinations for patients, and a quicker, less costly and more accurate decision support for the physician.

We present sktime -- a new scikit-learn compatible Python library with a unified interface for machine learning with time series. Time series data gives rise to various distinct but closely related learning tasks, such as forecasting and time series classification, many of which can be solved by reducing them to related simpler tasks. We discuss the main rationale for creating a unified interface, including reduction, as well as the design of sktime's core API, supported by a clear overview of common time series tasks and reduction approaches.

This paper studies the problem of predicting the distribution over multiple possible future paths of people as they move through various visual scenes. We make two main contributions. The first contribution is a new dataset, created in a realistic 3D simulator, which is based on real world trajectory data, and then extrapolated by human annotators to achieve different latent goals. This provides the first benchmark for quantitative evaluation of the models to predict multi-future trajectories. The second contribution is a new model to generate multiple plausible future trajectories, which contains novel designs of using multi-scale location encodings and convolutional RNNs over graphs. We refer to our model as Multiverse. We show that our model achieves the best results on our dataset, as well as on the real-world VIRAT/ActEV dataset (which just contains one possible future).

Clustering is a fundamental task in data analysis. Recently, deep clustering, which derives inspiration primarily from deep learning approaches, achieves state-of-the-art performance and has attracted considerable attention. Current deep clustering methods usually boost the clustering results by means of the powerful representation ability of deep learning, e.g., autoencoder, suggesting that learning an effective representation for clustering is a crucial requirement. The strength of deep clustering methods is to extract the useful representations from the data itself, rather than the structure of data, which receives scarce attention in representation learning. Motivated by the great success of Graph Convolutional Network (GCN) in encoding the graph structure, we propose a Structural Deep Clustering Network (SDCN) to integrate the structural information into deep clustering. Specifically, we design a delivery operator to transfer the representations learned by autoencoder to the corresponding GCN layer, and a dual self-supervised mechanism to unify these two different deep neural architectures and guide the update of the whole model. In this way, the multiple structures of data, from low-order to high-order, are naturally combined with the multiple representations learned by autoencoder. Furthermore, we theoretically analyze the delivery operator, i.e., with the delivery operator, GCN improves the autoencoder-specific representation as a high-order graph regularization constraint and autoencoder helps alleviate the over-smoothing problem in GCN. Through comprehensive experiments, we demonstrate that our propose model can consistently perform better over the state-of-the-art techniques.

Time series and signals are attracting more attention across statistics, machine learning and pattern recognition as it appears widely in the industry especially in sensor and IoT related research and applications, but few advances has been achieved in effective time series visual analytics and interaction due to its temporal dimensionality and complex dynamics. Inspired by recent effort on using network metrics to characterize time series for classification, we present an approach to visualize time series as complex networks based on the first order Markov process in its temporal ordering. In contrast to the classical bar charts, line plots and other statistics based graph, our approach delivers more intuitive visualization that better preserves both the temporal dependency and frequency structures. It provides a natural inverse operation to map the graph back to raw signals, making it possible to use graph statistics to characterize time series for better visual exploration and statistical analysis. Our experimental results suggest the effectiveness on various tasks such as pattern discovery and classification on both synthetic and the real time series and sensor data.

Multi-source transfer learning has been proven effective when within-target labeled data is scarce. Previous work focuses primarily on exploiting domain similarities and assumes that source domains are richly or at least comparably labeled. While this strong assumption is never true in practice, this paper relaxes it and addresses challenges related to sources with diverse labeling volume and diverse reliability. The first challenge is combining domain similarity and source reliability by proposing a new transfer learning method that utilizes both source-target similarities and inter-source relationships. The second challenge involves pool-based active learning where the oracle is only available in source domains, resulting in an integrated active transfer learning framework that incorporates distribution matching and uncertainty sampling. Extensive experiments on synthetic and two real-world datasets clearly demonstrate the superiority of our proposed methods over several baselines including state-of-the-art transfer learning methods.

Internet of Things (IoT)-based indoor localization has gained significant popularity recently to satisfy the ever-increasing requirements of indoor Location-based Services (LBS). In this context, Inertial Measurement Unit (IMU)-based localization is of interest as it provides a scalable solution independent of any proprietary sensors/modules. Existing IMU-based methodologies, however, are mainly developed based on statistical heading and step length estimation techniques that suffer from cumulative error issues and have extensive computational time requirements limiting their application for real-time indoor positioning. To address the aforementioned issues, we propose the Online Dynamic Window (ODW)-assisted two-stage Long Short Term Memory (LSTM) localization framework. Three ODWs are proposed, where the first model uses a Natural Language Processing (NLP)-inspired Dynamic Window (DW) approach, which significantly reduces the required computational time. The second framework is developed based on a Signal Processing Dynamic Windowing (SP-DW) approach to further reduce the required processing time of the two-stage LSTM-based model. The third ODW, referred to as the SP-NLP, combines the first two windowing mechanisms to further improve the overall achieved accuracy. Compared to the traditional LSTM-based positioning approaches, which suffer from either high tensor computation requirements or low accuracy, the proposed ODW-assisted models can perform indoor localization in a near-real time fashion with high accuracy. Performances of the proposed ODW-assisted models are evaluated based on a real Pedestrian Dead Reckoning (PDR) dataset. The results illustrate potentials of the proposed ODW-assisted techniques in achieving high classification accuracy with significantly reduced computational time, making them applicable for near real-time implementations.

Deep neural networks are capable of training fast and generalizing well within many domains. Despite their promising performance, deep networks have shown sensitivities to perturbations of their inputs (e.g., adversarial examples) and their learned feature representations are often difficult to interpret, raising concerns about their true capability and trustworthiness. Recent work in adversarial training, a form of robust optimization in which the model is optimized against adversarial examples, demonstrates the ability to improve performance sensitivities to perturbations and yield feature representations that are more interpretable. Adversarial training, however, comes with an increased computational cost over that of standard (i.e., nonrobust) training, rendering it impractical for use in large-scale problems. Recent work suggests that a fast approximation to adversarial training shows promise for reducing training time and maintaining robustness in the presence of perturbations bounded by the infinity norm. In this work, we demonstrate that this approach extends to the Euclidean norm and preserves the human-aligned feature representations that are common for robust models. Additionally, we show that using a distributed training scheme can further reduce the time to train robust deep networks. Fast adversarial training is a promising approach that will provide increased security and explainability in machine learning applications for which robust optimization was previously thought to be impractical.

For ego-motion estimation, the feature representation of the scenes is crucial. Previous methods indicate that both the low-level and semantic feature-based methods can achieve promising results. Therefore, the incorporation of hierarchical feature representation may benefit from both methods. From this perspective, we propose a novel direct feature odometry framework, named DFO, for depth estimation and hierarchical feature representation learning from monocular videos. By exploiting the metric distance, our framework is able to learn the hierarchical feature representation without supervision. The pose is obtained with a coarse-to-fine approach from high-level to low-level features in enlarged feature maps. The pixel-level attention mask can be self-learned to provide the prior information. In contrast to the previous methods, our proposed method calculates the camera motion with a direct method rather than regressing the ego-motion from the pose network. With this approach, the consistency of the scale factor of translation can be constrained. Additionally, the proposed method is thus compatible with the traditional SLAM pipeline. Experiments on the KITTI dataset demonstrate the effectiveness of our method.

This paper considers minimax optimization $\min_x \max_y f(x, y)$ in the challenging setting where $f$ can be both nonconvex in $x$ and nonconcave in $y$. Though such optimization problems arise in many machine learning paradigms including training generative adversarial networks (GANs) and adversarially robust models, many fundamental issues remain in theory, such as the absence of efficiently computable optimality notions, and cyclic or diverging behavior of existing algorithms. Our framework sprouts from the practical consideration that under a computational budget, the max-player can not fully maximize $f(x,\cdot)$ since nonconcave maximization is NP-hard in general. So, we propose a new algorithm for the min-player to play against smooth algorithms deployed by the adversary (i.e., the max-player) instead of against full maximization. Our algorithm is guaranteed to make monotonic progress (thus having no limit cycles), and to find an appropriate "stationary point" in a polynomial number of iterations. Our framework covers practical settings where the smooth algorithms deployed by the adversary are multi-step stochastic gradient ascent, and its accelerated version. We further provide complementing experiments that confirm our theoretical findings and demonstrate the effectiveness of the proposed approach in practice.

The UCR Time Series Archive - introduced in 2002, has become an important resource in the time series data mining community, with at least one thousand published papers making use of at least one data set from the archive. The original incarnation of the archive had sixteen data sets but since that time, it has gone through periodic expansions. The last expansion took place in the summer of 2015 when the archive grew from 45 to 85 data sets. This paper introduces and will focus on the new data expansion from 85 to 128 data sets. Beyond expanding this valuable resource, this paper offers pragmatic advice to anyone who may wish to evaluate a new algorithm on the archive. Finally, this paper makes a novel and yet actionable claim: of the hundreds of papers that show an improvement over the standard baseline (1-nearest neighbor classification), a large fraction may be mis-attributing the reasons for their improvement. Moreover, they may have been able to achieve the same improvement with a much simpler modification, requiring just a single line of code.

Most of previous machine learning algorithms are proposed based on the i.i.d. hypothesis. However, this ideal assumption is often violated in real applications, where selection bias may arise between training and testing process. Moreover, in many scenarios, the testing data is not even available during the training process, which makes the traditional methods like transfer learning infeasible due to their need on prior of test distribution. Therefore, how to address the agnostic selection bias for robust model learning is of paramount importance for both academic research and real applications. In this paper, under the assumption that causal relationships among variables are robust across domains, we incorporate causal technique into predictive modeling and propose a novel Causally Regularized Logistic Regression (CRLR) algorithm by jointly optimize global confounder balancing and weighted logistic regression. Global confounder balancing helps to identify causal features, whose causal effect on outcome are stable across domains, then performing logistic regression on those causal features constructs a robust predictive model against the agnostic bias. To validate the effectiveness of our CRLR algorithm, we conduct comprehensive experiments on both synthetic and real world datasets. Experimental results clearly demonstrate that our CRLR algorithm outperforms the state-of-the-art methods, and the interpretability of our method can be fully depicted by the feature visualization.

AlphaStar, the AI that reaches GrandMaster level in StarCraft II, is a remarkable milestone demonstrating what deep reinforcement learning can achieve in complex Real-Time Strategy (RTS) games. However, the complexities of the game, algorithms and systems, and especially the tremendous amount of computation needed are big obstacles for the community to conduct further research in this direction. We propose a deep reinforcement learning agent, StarCraft Commander (SCC). With order of magnitude less computation, it demonstrates top human performance defeating GrandMaster players in test matches and top professional players in a live event. Moreover, it shows strong robustness to various human strategies and discovers novel strategies unseen from human plays. In this paper, we will share the key insights and optimizations on efficient imitation learning and reinforcement learning for StarCraft II full game.

Cross-domain object detection is challenging, because object detection model is often vulnerable to data variance, especially to the considerable domain shift between two distinctive domains. In this paper, we propose a new Unbiased Mean Teacher (UMT) model for cross-domain object detection. We reveal that there often exists a considerable model bias for the simple mean teacher (MT) model in cross-domain scenarios, and eliminate the model bias with several simple yet highly effective strategies. In particular, for the teacher model, we propose a cross-domain distillation method for MT to maximally exploit the expertise of the teacher model. Moreover, for the student model, we alleviate its bias by augmenting training samples with pixel-level adaptation. Finally, for the teaching process, we employ an out-of-distribution estimation strategy to select samples that most fit the current model to further enhance the cross-domain distillation process. By tackling the model bias issue with these strategies, our UMT model achieves mAPs of 44.1%, 58.1%, 41.7%, and 43.1% on benchmark datasets Clipart1k, Watercolor2k, Foggy Cityscapes, and Cityscapes, respectively, which outperforms the existing state-of-the-art results in notable margins. Our implementation is available at https://github.com/kinredon/umt.

Fair representation learning aims to encode invariant representation with respect to the protected attribute, such as gender or age. In this paper, we design Fairness-aware Disentangling Variational AutoEncoder (FD-VAE) for fair representation learning. This network disentangles latent space into three subspaces with a decorrelation loss that encourages each subspace to contain independent information: 1) target attribute information, 2) protected attribute information, 3) mutual attribute information. After the representation learning, this disentangled representation is leveraged for fairer downstream classification by excluding the subspace with the protected attribute information. We demonstrate the effectiveness of our model through extensive experiments on CelebA and UTK Face datasets. Our method outperforms the previous state-of-the-art method by large margins in terms of equal opportunity and equalized odds.

Intensive care clinicians need reliable clinical practice tools to preempt unexpected critical events that might harm their patients in intensive care units (ICU), to pre-plan timely interventions, and to keep the patient's family well informed. The conventional statistical models are built by curating only a limited number of key variables, which means a vast unknown amount of potentially precious data remains unused. Deep learning models (DLMs) can be leveraged to learn from large complex datasets and construct predictive clinical tools. This retrospective study was performed using 42,818 hospital admissions involving 35,348 patients, which is a subset of the MIMIC-III dataset. Natural language processing (NLP) techniques were applied to build DLMs to predict in-hospital mortality (IHM) and length of stay >=7 days (LOS). Over 75 million events across multiple data sources were processed, resulting in over 355 million tokens. DLMs for predicting IHM using data from all sources (AS) and chart data (CS) achieved an AUC-ROC of 0.9178 and 0.9029, respectively, and PR-AUC of 0.6251 and 0.5701, respectively. DLMs for predicting LOS using AS and CS achieved an AUC-ROC of 0.8806 and 0.8642, respectively, and PR-AUC of 0.6821 and 0.6575, respectively. The observed AUC-ROC difference between models was found to be significant for both IHM and LOS at p=0.05. The observed PR-AUC difference between the models was found to be significant for IHM and statistically insignificant for LOS at p=0.05. In this study, deep learning models were constructed using data combined from a variety of sources in Electronic Health Records (EHRs) such as chart data, input and output events, laboratory values, microbiology events, procedures, notes, and prescriptions. It is possible to predict in-hospital mortality with much better confidence and higher reliability from models built using all sources of data.

This paper explores conditional image generation with a One-Vs-All classifier based on the Generative Adversarial Networks (GANs). Instead of the real/fake discriminator used in vanilla GANs, we propose to extend the discriminator to a One-Vs-All classifier (GAN-OVA) that can distinguish each input data to its category label. Specifically, we feed certain additional information as conditions to the generator and take the discriminator as a One-Vs-All classifier to identify each conditional category. Our model can be applied to different divergence or distances used to define the objective function, such as Jensen-Shannon divergence and Earth-Mover (or called Wasserstein-1) distance. We evaluate GAN-OVAs on MNIST and CelebA-HQ datasets, and the experimental results show that GAN-OVAs make progress toward stable training over regular conditional GANs. Furthermore, GAN-OVAs effectively accelerate the generation process of different classes and improves generation quality.

Motivated by the pursuit of a systematic computational and algorithmic understanding of Generative Adversarial Networks (GANs), we present a simple yet unified non-asymptotic local convergence theory for smooth two-player games, which subsumes several discrete-time gradient-based saddle point dynamics. The analysis reveals the surprising nature of the off-diagonal interaction term as both a blessing and a curse. On the one hand, this interaction term explains the origin of the slow-down effect in the convergence of Simultaneous Gradient Ascent (SGA) to stable Nash equilibria. On the other hand, for the unstable equilibria, exponential convergence can be proved thanks to the interaction term, for four modified dynamics proposed to stabilize GAN training: Optimistic Mirror Descent (OMD), Consensus Optimization (CO), Implicit Updates (IU) and Predictive Method (PM). The analysis uncovers the intimate connections among these stabilizing techniques, and provides detailed characterization on the choice of learning rate. As a by-product, we present a new analysis for OMD proposed in Daskalakis, Ilyas, Syrgkanis, and Zeng [2017] with improved rates.

Cryogenic electron microscopy (cryo-EM) has become an enabling technology in drug discovery and in understanding molecular bases of disease by producing near-atomic resolution (less than 0.4 nm) 3D reconstructions of biological macromolecules. The imaging process required for 3D reconstructions involves a highly iterative and empirical screening process, starting with the acquisition of low magnification images of the cryo-EM grids. These images are inspected for squares that are likely to contain useful molecular signals. Potentially useful squares within the grid are then imaged at progressively higher magnifications, with the goal of identifying sub-micron areas within circular holes (bounded by the squares) for imaging at high magnification. This arduous, multi-step data acquisition process represents a bottleneck for obtaining a high throughput data collection. Here, we focus on automating the early decision making for the microscope operator, scoring low magnification images of squares, and proposing the first deep learning framework, XCryoNet, for automated cryo-EM grid screening. XCryoNet is a semi-supervised, attention-guided deep learning approach that provides explainable scoring of automatically extracted square images using limited amounts of labeled data. Results show up to 8% and 37% improvements over a fully supervised and a no-attention solution, respectively, when labeled data is scarce.

Forecasting of multivariate time-series is an important problem that has applications in traffic management, cellular network configuration, and quantitative finance. A special case of the problem arises when there is a graph available that captures the relationships between the time-series. In this paper we propose a novel learning architecture that achieves performance competitive with or better than the best existing algorithms, without requiring knowledge of the graph. The key element of our proposed architecture is the learnable fully connected hard graph gating mechanism that enables the use of the state-of-the-art and highly computationally efficient fully connected time-series forecasting architecture in traffic forecasting applications. Experimental results for two public traffic network datasets illustrate the value of our approach, and ablation studies confirm the importance of each element of the architecture. The code is available here: https://github.com/boreshkinai/fc-gaga.

For early breast cancer detection, regular screening with mammography imaging is recommended. Routinary examinations result in datasets with a predominant amount of negative samples. A potential solution to such class-imbalance is joining forces across multiple institutions. Developing a collaborative computer-aided diagnosis system is challenging in different ways. Patient privacy and regulations need to be carefully respected. Data across institutions may be acquired from different devices or imaging protocols, leading to heterogeneous non-IID data. Also, for learning-based methods, new optimization strategies working on distributed data are required. Recently, federated learning has emerged as an effective tool for collaborative learning. In this setting, local models perform computation on their private data to update the global model. The order and the frequency of local updates influence the final global model. Hence, the order in which samples are locally presented to the optimizers plays an important role. In this work, we define a memory-aware curriculum learning method for the federated setting. Our curriculum controls the order of the training samples paying special attention to those that are forgotten after the deployment of the global model. Our approach is combined with unsupervised domain adaptation to deal with domain shift while preserving data privacy. We evaluate our method with three clinical datasets from different vendors. Our results verify the effectiveness of federated adversarial learning for the multi-site breast cancer classification. Moreover, we show that our proposed memory-aware curriculum method is beneficial to further improve classification performance. Our code is publicly available at: https://github.com/ameliajimenez/curriculum-federated-learning.

We present CROP (Central Roundish Object Painter), which identifies and paints the object at the center of an RGB image. Primarily CROP works for roundish fruits in various illumination conditions, but surprisingly, it could also deal with images of other organic or inorganic materials, or ones by optical and electron microscopes, although CROP was trained solely by 172 images of fruits. The method involves image segmentation by deep learning, and the architecture of the neural network is a deeper version of the original U-Net. This technique could provide us with a means of automatically collecting statistical data of fruit growth in farms. As an example, we describe our experiment of processing 510 time series photos automatically to collect the data on the size and the position of the target fruit. Our trained neural network CROP and the above automatic programs are available on GitHub with user-friendly interface programs.

Benefiting from the powerful expressive capability of graphs, graph-based approaches have achieved impressive performance in various biomedical applications. Most existing methods tend to define the adjacency matrix among samples manually based on meta-features, and then obtain the node embeddings for downstream tasks by Graph Representation Learning (GRL). However, it is not easy for these approaches to generalize to unseen samples. Meanwhile, the complex correlation between modalities is also ignored. As a result, these factors inevitably yield the inadequacy of providing valid information about the patient's condition for a reliable diagnosis. In this paper, we propose an end-to-end Multimodal Graph Learning framework (MMGL) for disease prediction. To effectively exploit the rich information across multi-modality associated with diseases, amodal-attentional multi-modal fusion is proposed to integrate the features of each modality by leveraging the correlation and complementarity between the modalities. Furthermore, instead of defining the adjacency matrix manually as existing methods, the latent graph structure can be captured through a novel way of adaptive graph learning. It could be jointly optimized with the prediction model, thus revealing the intrinsic connections among samples. Unlike the previous transductive methods, our model is also applicable to the scenario of inductive learning for those unseen data. An extensive group of experiments on two disease prediction problems is then carefully designed and presented, demonstrating that MMGL obtains more favorable performances. In addition, we also visualize and analyze the learned graph structure to provide more reliable decision support for doctors in real medical applications and inspiration for disease research.

We propose a new neural sequence model training method in which the objective function is defined by $\alpha$-divergence. We demonstrate that the objective function generalizes the maximum-likelihood (ML)-based and reinforcement learning (RL)-based objective functions as special cases (i.e., ML corresponds to $\alpha \to 0$ and RL to $\alpha \to1$). We also show that the gradient of the objective function can be considered a mixture of ML- and RL-based objective gradients. The experimental results of a machine translation task show that minimizing the objective function with $\alpha > 0$ outperforms $\alpha \to 0$, which corresponds to ML-based methods.

Recurrent neural networks (RNNs) have shown the ability to improve scene parsing through capturing long-range dependencies among image units. In this paper, we propose dense RNNs for scene labeling by exploring various long-range semantic dependencies among image units. Different from existing RNN based approaches, our dense RNNs are able to capture richer contextual dependencies for each image unit by enabling immediate connections between each pair of image units, which significantly enhances their discriminative power. Besides, to select relevant dependencies and meanwhile to restrain irrelevant ones for each unit from dense connections, we introduce an attention model into dense RNNs. The attention model allows automatically assigning more importance to helpful dependencies while less weight to unconcerned dependencies. Integrating with convolutional neural networks (CNNs), we develop an end-to-end scene labeling system. Extensive experiments on three large-scale benchmarks demonstrate that the proposed approach can improve the baselines by large margins and outperform other state-of-the-art algorithms.

Estimates of image gradients play a ubiquitous role in image segmentation and classification problems since gradients directly relate to the boundaries or the edges of a scene. This paper proposes an unified approach to gradient estimation based on fractional calculus that is computationally cheap and readily applicable to any existing algorithm that relies on image gradients. We show experiments on edge detection and image segmentation on the Stanford Backgrounds Dataset where these improved local gradients outperforms state of the art, achieving a performance of 79.2% average accuracy.

Stifle joint issues are a major cause of lameness in dogs and it can be a significant marker for various forms of diseases or injuries. A known Tibial Plateau Angle (TPA) helps in the reduction of the diagnosis time of the cause. With the state of the art object detection algorithm YOLO, and its variants, this paper delves into identifying joints, their centroids and other regions of interest to draw multiple line axes and finally calculating the TPA. The methods investigated predicts successfully the TPA within the normal range for 80 percent of the images.

Three-dimensional medical image segmentation is one of the most important problems in medical image analysis and plays a key role in downstream diagnosis and treatment. Recent years, deep neural networks have made groundbreaking success in medical image segmentation problem. However, due to the high variance in instrumental parameters, experimental protocols, and subject appearances, the generalization of deep learning models is often hindered by the inconsistency in medical images generated by different machines and hospitals. In this work, we present StyleSegor, an efficient and easy-to-use strategy to alleviate this inconsistency issue. Specifically, neural style transfer algorithm is applied to unlabeled data in order to minimize the differences in image properties including brightness, contrast, texture, etc. between the labeled and unlabeled data. We also apply probabilistic adjustment on the network output and integrate multiple predictions through ensemble learning. On a publicly available whole heart segmentation benchmarking dataset from MICCAI HVSMR 2016 challenge, we have demonstrated an elevated dice accuracy surpassing current state-of-the-art method and notably, an improvement of the total score by 29.91\%. StyleSegor is thus corroborated to be an accurate tool for 3D whole heart segmentation especially on highly inconsistent data, and is available at https://github.com/horsepurve/StyleSegor.

Image clustering is to group a set of images into disjoint clusters in a way that images in the same cluster are more similar to each other than to those in other clusters, which is an unsupervised or semi-supervised learning process. It is a crucial and challenging task in machine learning and computer vision. The performances of existing image clustering methods have close relations with features used for clustering, even if unsupervised coding based methods have improved the performances a lot. To reduce the effect of clustering features, we propose a feature-weak-relevant method for image clustering. The proposed method converts an unsupervised clustering process into an alternative iterative process of unsupervised learning and transfer learning. The clustering process firstly starts up from handcrafted features based image clustering to estimate an initial label for every image, and secondly use a proposed sampling strategy to choose images with reliable labels to feed a transfer-learning model to learn representative features that can be used for next round of unsupervised learning. In this manner, image clustering is iteratively optimized. What's more, the handcrafted features are used to boot up the clustering process, and just have a little effect on the final performance; therefore, the proposed method is feature-weak-relevant. Experimental results on six kinds of public available datasets show that the proposed method outperforms state of the art methods and depends less on the employed features at the same time.

Recently, attention-based encoder-decoder models have been used extensively in image captioning. Yet there is still great difficulty for the current methods to achieve deep image understanding. In this work, we argue that such understanding requires visual attention to correlated image regions and semantic attention to coherent attributes of interest. Based on the Transformer, to perform effective attention, we explore image captioning from a cross-modal perspective and propose the Global-and-Local Information Exploring-and-Distilling approach that explores and distills the source information in vision and language. It globally provides the aspect vector, a spatial and relational representation of images based on caption contexts, through the extraction of salient region groupings and attribute collocations, and locally extracts the fine-grained regions and attributes in reference to the aspect vector for word selection. Our Transformer-based model achieves a CIDEr score of 129.3 in offline COCO evaluation on the COCO testing set with remarkable efficiency in terms of accuracy, speed, and parameter budget.

While existing makeup style transfer models perform an image synthesis whose results cannot be explicitly controlled, the ability to modify makeup color continuously is a desirable property for virtual try-on applications. We propose a new formulation for the makeup style transfer task, with the objective to learn a color controllable makeup style synthesis. We introduce CA-GAN, a generative model that learns to modify the color of specific objects (e.g. lips or eyes) in the image to an arbitrary target color while preserving background. Since color labels are rare and costly to acquire, our method leverages weakly supervised learning for conditional GANs. This enables to learn a controllable synthesis of complex objects, and only requires a weak proxy of the image attribute that we desire to modify. Finally, we present for the first time a quantitative analysis of makeup style transfer and color control performance.

Multi-agent deep reinforcement learning (MARL) suffers from a lack of commonly-used evaluation tasks and criteria, making comparisons between approaches difficult. In this work, we consistently evaluate and compare three different classes of MARL algorithms (independent learning, centralised multi-agent policy gradient, value decomposition) in a diverse range of cooperative multi-agent learning tasks. Our experiments serve as a reference for the expected performance of algorithms across different learning tasks, and we provide insights regarding the effectiveness of different learning approaches. We open-source EPyMARL, which extends the PyMARL codebase~\citep{samvelyan19smac} to include additional algorithms and allow for flexible configuration of algorithm implementation details such as parameter sharing. Finally, we open-source two environments for multi-agent research which focus on coordination under sparse rewards.

While semi-supervised learning (SSL) algorithms provide an efficient way to make use of both labelled and unlabelled data, they generally struggle when the number of annotated samples is very small. In this work, we consider the problem of SSL multi-class classification with very few labelled instances. We introduce two key ideas. The first is a simple but effective one: we leverage the power of transfer learning among different tasks and self-supervision to initialize a good representation of the data without making use of any label. The second idea is a new algorithm for SSL that can exploit well such a pre-trained representation. The algorithm works by alternating two phases, one fitting the labelled points and one fitting the unlabelled ones, with carefully-controlled information flow between them. The benefits are greatly reducing overfitting of the labelled data and avoiding issue with balancing labelled and unlabelled losses during training. We show empirically that this method can successfully train competitive models with as few as 10 labelled data points per class. More in general, we show that the idea of bootstrapping features using self-supervised learning always improves SSL on standard benchmarks. We show that our algorithm works increasingly well compared to other methods when refining from other tasks or datasets.

Real-world data is often unbalanced and long-tailed, but deep models struggle to recognize rare classes in the presence of frequent classes. To address unbalanced data, most studies try balancing the data, the loss, or the classifier to reduce classification bias towards head classes. Far less attention has been given to the latent representations learned with unbalanced data. We show that the feature extractor part of deep networks suffers greatly from this bias. We propose a new loss based on robustness theory, which encourages the model to learn high-quality representations for both head and tail classes. While the general form of the robustness loss may be hard to compute, we further derive an easy-to-compute upper bound that can be minimized efficiently. This procedure reduces representation bias towards head classes in the feature space and achieves new SOTA results on CIFAR100-LT, ImageNet-LT, and iNaturalist long-tail benchmarks. We find that training with robustness increases recognition accuracy of tail classes while largely maintaining the accuracy of head classes. The new robustness loss can be combined with various classifier balancing techniques and can be applied to representations at several layers of the deep model.

The application of deep learning to medical image segmentation has been hampered due to the lack of abundant pixel-level annotated data. Few-shot Semantic Segmentation (FSS) is a promising strategy for breaking the deadlock. However, a high-performing FSS model still requires sufficient pixel-level annotated classes for training to avoid overfitting, which leads to its performance bottleneck in medical image segmentation due to the unmet need for annotations. Thus, semi-supervised FSS for medical images is accordingly proposed to utilize unlabeled data for further performance improvement. Nevertheless, existing semi-supervised FSS methods has two obvious defects: (1) neglecting the relationship between the labeled and unlabeled data; (2) using unlabeled data directly for end-to-end training leads to degenerated representation learning. To address these problems, we propose a novel semi-supervised FSS framework for medical image segmentation. The proposed framework employs Poisson learning for modeling data relationship and propagating supervision signals, and Spatial Consistency Calibration for encouraging the model to learn more coherent representations. In this process, unlabeled samples do not involve in end-to-end training, but provide supervisory information for query image segmentation through graph-based learning. We conduct extensive experiments on three medical image segmentation datasets (i.e. ISIC skin lesion segmentation, abdominal organs segmentation for MRI and abdominal organs segmentation for CT) to demonstrate the state-of-the-art performance and broad applicability of the proposed framework.

We propose a novel method to use both audio and a low-resolution image to perform extreme face super-resolution (a 16x increase of the input size). When the resolution of the input image is very low (e.g., 8x8 pixels), the loss of information is so dire that important details of the original identity have been lost and audio can aid the recovery of a plausible high-resolution image. In fact, audio carries information about facial attributes, such as gender and age. To combine the aural and visual modalities, we propose a method to first build the latent representations of a face from the lone audio track and then from the lone low-resolution image. We then train a network to fuse these two representations. We show experimentally that audio can assist in recovering attributes such as the gender, the age and the identity, and thus improve the correctness of the high-resolution image reconstruction process. Our procedure does not make use of human annotation and thus can be easily trained with existing video datasets. Moreover, we show that our model builds a factorized representation of images and audio as it allows one to mix low-resolution images and audio from different videos and to generate realistic faces with semantically meaningful combinations.

Constructing appropriate representations of molecules lies at the core of numerous tasks such as material science, chemistry and drug designs. Recent researches abstract molecules as attributed graphs and employ graph neural networks (GNN) for molecular representation learning, which have made remarkable achievements in molecular graph modeling. Albeit powerful, current models either are based on local aggregation operations and thus miss higher-order graph properties or focus on only node information without fully using the edge information. For this sake, we propose a Communicative Message Passing Transformer (CoMPT) neural network to improve the molecular graph representation by reinforcing message interactions between nodes and edges based on the Transformer architecture. Unlike the previous transformer-style GNNs that treat molecules as fully connected graphs, we introduce a message diffusion mechanism to leverage the graph connectivity inductive bias and reduce the message enrichment explosion. Extensive experiments demonstrated that the proposed model obtained superior performances (around 4$\%$ on average) against state-of-the-art baselines on seven chemical property datasets (graph-level tasks) and two chemical shift datasets (node-level tasks). Further visualization studies also indicated a better representation capacity achieved by our model.

In the context of visual navigation, the capacity to map a novel environment is necessary for an agent to exploit its observation history in the considered place and efficiently reach known goals. This ability can be associated with spatial reasoning, where an agent is able to perceive spatial relationships and regularities, and discover object characteristics. In classical Reinforcement Learning (RL) setups, this capacity is learned from reward alone. We introduce supplementary supervision in the form of auxiliary tasks designed to favor the emergence of spatial perception capabilities in agents trained for a goal-reaching downstream objective. We show that learning to estimate metrics quantifying the spatial relationships between an agent at a given location and a goal to reach has a high positive impact in Multi-Object Navigation settings. Our method significantly improves the performance of different baseline agents, that either build an explicit or implicit representation of the environment, even matching the performance of incomparable oracle agents taking ground-truth maps as input.

Multi-View Stereo (MVS) is a core task in 3D computer vision. With the surge of novel deep learning methods, learned MVS has surpassed the accuracy of classical approaches, but still relies on building a memory intensive dense cost volume. Novel View Synthesis (NVS) is a parallel line of research and has recently seen an increase in popularity with Neural Radiance Field (NeRF) models, which optimize a per scene radiance field. However, NeRF methods do not generalize to novel scenes and are slow to train and test. We propose to bridge the gap between these two methodologies with a novel network that can recover 3D scene geometry as a distance function, together with high-resolution color images. Our method uses only a sparse set of images as input and can generalize well to novel scenes. Additionally, we propose a coarse-to-fine sphere tracing approach in order to significantly increase speed. We show on various datasets that our method reaches comparable accuracy to per-scene optimized methods while being able to generalize and running significantly faster.

We consider the problem of batch multi-task reinforcement learning with observed context descriptors, motivated by its application to personalized medical treatment. In particular, we study two general classes of learning algorithms: direct policy learning (DPL), an imitation-learning based approach which learns from expert trajectories, and model-based learning. First, we derive sample complexity bounds for DPL, and then show that model-based learning from expert actions can, even with a finite model class, be impossible. After relaxing the conditions under which the model-based approach is expected to learn by allowing for greater coverage of state-action space, we provide sample complexity bounds for model-based learning with finite model classes, showing that there exist model classes with sample complexity exponential in their statistical complexity. We then derive a sample complexity upper bound for model-based learning based on a measure of concentration of the data distribution. Our results give formal justification for imitation learning over model-based learning in this setting.

Self-supervised monocular depth estimation presents a powerful method to obtain 3D scene information from single camera images, which is trainable on arbitrary image sequences without requiring depth labels, e.g., from a LiDAR sensor. In this work we present a new self-supervised semantically-guided depth estimation (SGDepth) method to deal with moving dynamic-class (DC) objects, such as moving cars and pedestrians, which violate the static-world assumptions typically made during training of such models. Specifically, we propose (i) mutually beneficial cross-domain training of (supervised) semantic segmentation and self-supervised depth estimation with task-specific network heads, (ii) a semantic masking scheme providing guidance to prevent moving DC objects from contaminating the photometric loss, and (iii) a detection method for frames with non-moving DC objects, from which the depth of DC objects can be learned. We demonstrate the performance of our method on several benchmarks, in particular on the Eigen split, where we exceed all baselines without test-time refinement.

Inspired by the recent developments in computer vision, low-rank and structured sparse matrix decomposition can be potentially be used for extract moving objects in satellite videos. This set of approaches seeks for rank minimization on the background that typically requires batch-based optimization over a sequence of frames, which causes delays in processing and limits their applications. To remedy this delay, we propose an Online Low-rank and Structured Sparse Decomposition (O-LSD). O-LSD reformulates the batch-based low-rank matrix decomposition with the structured sparse penalty to its equivalent frame-wise separable counterpart, which then defines a stochastic optimization problem for online subspace basis estimation. In order to promote online processing, O-LSD conducts the foreground and background separation and the subspace basis update alternatingly for every frame in a video. We also show the convergence of O-LSD theoretically. Experimental results on two satellite videos demonstrate the performance of O-LSD in term of accuracy and time consumption is comparable with the batch-based approaches with significantly reduced delay in processing.

Street-to-aerial image geo-localization, which matches a query street-view image to the GPS-tagged aerial images in a reference set, has attracted increasing attention recently. In this paper, we revisit this problem and point out the ignored issue about image alignment information. We show that the performance of a simple Siamese network is highly dependent on the alignment setting and the comparison of previous works can be unfair if they have different assumptions. Instead of focusing on the feature extraction under the alignment assumption, we show that improvements in metric learning techniques significantly boost the performance regardless of the alignment. Without leveraging the alignment information, our pipeline outperforms previous works on both panorama and cropped datasets. Furthermore, we conduct visualization to help understand the learned model and the effect of alignment information using Grad-CAM. With our discovery on the approximate rotation-invariant activation maps, we propose a novel method to estimate the orientation/alignment between a pair of cross-view images with unknown alignment information. It achieves state-of-the-art results on the CVUSA dataset.

Carbon capture and storage (CCS) can aid decarbonization of the atmosphere to limit further global temperature increases. A framework utilizing unsupervised learning is used to generate a range of subsurface geologic volumes to investigate potential sites for long-term storage of carbon dioxide. Generative adversarial networks are used to create geologic volumes, with a further neural network used to sample the posterior distribution of a trained Generator conditional to sparsely sampled physical measurements. These generative models are further conditioned to historic dynamic fluid flow data through Bayesian inversion to improve the resolution of the forecast of the storage capacity of injected carbon dioxide.

This article introduces the solutions of the two champion teams, `MMfruit' for the detection track and `MMfruitSeg' for the segmentation track, in OpenImage Challenge 2019. It is commonly known that for an object detector, the shared feature at the end of the backbone is not appropriate for both classification and regression, which greatly limits the performance of both single stage detector and Faster RCNN \cite{ren2015faster} based detector. In this competition, we observe that even with a shared feature, different locations in one object has completely inconsistent performances for the two tasks. \textit{E.g. the features of salient locations are usually good for classification, while those around the object edge are good for regression.} Inspired by this, we propose the Decoupling Head (DH) to disentangle the object classification and regression via the self-learned optimal feature extraction, which leads to a great improvement. Furthermore, we adjust the soft-NMS algorithm to adj-NMS to obtain stable performance improvement. Finally, a well-designed ensemble strategy via voting the bounding box location and confidence is proposed. We will also introduce several training/inferencing strategies and a bag of tricks that give minor improvement. Given those masses of details, we train and aggregate 28 global models with various backbones, heads and 3+2 expert models, and achieves the 1st place on the OpenImage 2019 Object Detection Challenge on the both public and private leadboards. Given such good instance bounding box, we further design a simple instance-level semantic segmentation pipeline and achieve the 1st place on the segmentation challenge.

Federated learning (FL) offers a solution to train a global machine learning model while still maintaining data privacy, without needing access to data stored locally at the clients. However, FL suffers performance degradation when client data distribution is non-IID, and a longer training duration to combat this degradation may not necessarily be feasible due to communication limitations. To address this challenge, we propose a new adaptive training algorithm $\texttt{AdaFL}$, which comprises two components: (i) an attention-based client selection mechanism for a fairer training scheme among the clients; and (ii) a dynamic fraction method to balance the trade-off between performance stability and communication efficiency. Experimental results show that our $\texttt{AdaFL}$ algorithm outperforms the usual $\texttt{FedAvg}$ algorithm, and can be incorporated to further improve various state-of-the-art FL algorithms, with respect to three aspects: model accuracy, performance stability, and communication efficiency.

Segmentation of colorectal cancerous regions from 3D Magnetic Resonance (MR) images is a crucial procedure for radiotherapy which conventionally requires accurate delineation of tumour boundaries at an expense of labor, time and reproducibility. While deep learning based methods serve good baselines in 3D image segmentation tasks, small applicable patch size limits effective receptive field and degrades segmentation performance. In addition, Regions of interest (RoIs) localization from large whole volume 3D images serves as a preceding operation that brings about multiple benefits in terms of speed, target completeness, reduction of false positives. Distinct from sliding window or non-joint localization-segmentation based models, we propose a novel multitask framework referred to as 3D RoI-aware U-Net (3D RU-Net), for RoI localization and in-region segmentation where the two tasks share one backbone encoder network. With the region proposals from the encoder, we crop multi-level RoI in-region features from the encoder to form a GPU memory-efficient decoder for detailpreserving segmentation and therefore enlarged applicable volume size and effective receptive field. To effectively train the model, we designed a Dice formulated loss function for the global-to-local multi-task learning procedure. Based on the efficiency gains, we went on to ensemble models with different receptive fields to achieve even higher performance costing minor extra computational expensiveness. Extensive experiments were conducted on 64 cancerous cases with a four-fold cross-validation, and the results showed significant superiority in terms of accuracy and efficiency over conventional frameworks. In conclusion, the proposed method has a huge potential for extension to other 3D object segmentation tasks from medical images due to its inherent generalizability. The code for the proposed method is publicly available.

Recently, contrastive learning has largely advanced the progress of unsupervised visual representation learning. Pre-trained on ImageNet, some self-supervised algorithms reported higher transfer learning performance compared to fully-supervised methods, seeming to deliver the message that human labels hardly contribute to learning transferrable visual features. In this paper, we defend the usefulness of semantic labels but point out that fully-supervised and self-supervised methods are pursuing different kinds of features. To alleviate this issue, we present a new algorithm named Supervised Contrastive Adjustment in Neighborhood (SCAN) that maximally prevents the semantic guidance from damaging the appearance feature embedding. In a series of downstream tasks, SCAN achieves superior performance compared to previous fully-supervised and self-supervised methods, and sometimes the gain is significant. More importantly, our study reveals that semantic labels are useful in assisting self-supervised methods, opening a new direction for the community.

Deep learning based approaches have been utilized to model and generate graphs subjected to different distributions recently. However, they are typically unsupervised learning based and unconditioned generative models or simply conditioned on the graph-level contexts, which are not associated with rich semantic node-level contexts. Differently, in this paper, we are interested in a novel problem named Time Series Conditioned Graph Generation: given an input multivariate time series, we aim to infer a target relation graph modeling the underlying interrelationships between time series with each node corresponding to each time series. For example, we can study the interrelationships between genes in a gene regulatory network of a certain disease conditioned on their gene expression data recorded as time series. To achieve this, we propose a novel Time Series conditioned Graph Generation-Generative Adversarial Networks (TSGG-GAN) to handle challenges of rich node-level context structures conditioning and measuring similarities directly between graphs and time series. Extensive experiments on synthetic and real-word gene regulatory networks datasets demonstrate the effectiveness and generalizability of the proposed TSGG-GAN.

Salient object detection has seen remarkable progress driven by deep learning techniques. However, most of deep learning based salient object detection methods are black-box in nature and lacking in interpretability. This paper proposes the first self-explanatory saliency detection network that explicitly exploits low- and high-level features for salient object detection. We demonstrate that such supportive clues not only significantly enhances performance of salient object detection but also gives better justified detection results. More specifically, we develop a multi-stage saliency encoder to extract multi-scale features which contain both low- and high-level saliency context. Dense short- and long-range connections are introduced to reuse these features iteratively. Benefiting from the direct access to low- and high-level features, the proposed saliency encoder can not only model the object context but also preserve the boundary. Furthermore, a self-explanatory generator is proposed to interpret how the proposed saliency encoder or other deep saliency models making decisions. The generator simulates the absence of interesting features by preventing these features from contributing to the saliency classifier and estimates the corresponding saliency prediction without these features. A comparison function, saliency explanation, is defined to measure the prediction changes between deep saliency models and corresponding generator. Through visualizing the differences, we can interpret the capability of different deep neural networks based saliency detection models and demonstrate that our proposed model indeed uses more reasonable structure for salient object detection. Extensive experiments on five popular benchmark datasets and the visualized saliency explanation demonstrate that the proposed method provides new state-of-the-art.

Policy distillation, which transfers a teacher policy to a student policy has achieved great success in challenging tasks of deep reinforcement learning. This teacher-student framework requires a well-trained teacher model which is computationally expensive. Moreover, the performance of the student model could be limited by the teacher model if the teacher model is not optimal. In the light of collaborative learning, we study the feasibility of involving joint intellectual efforts from diverse perspectives of student models. In this work, we introduce dual policy distillation(DPD), a student-student framework in which two learners operate on the same environment to explore different perspectives of the environment and extract knowledge from each other to enhance their learning. The key challenge in developing this dual learning framework is to identify the beneficial knowledge from the peer learner for contemporary learning-based reinforcement learning algorithms, since it is unclear whether the knowledge distilled from an imperfect and noisy peer learner would be helpful. To address the challenge, we theoretically justify that distilling knowledge from a peer learner will lead to policy improvement and propose a disadvantageous distillation strategy based on the theoretical results. The conducted experiments on several continuous control tasks show that the proposed framework achieves superior performance with a learning-based agent and function approximation without the use of expensive teacher models.

We propose a method for human activity recognition from RGB data that does not rely on any pose information during test time and does not explicitly calculate pose information internally. Instead, a visual attention module learns to predict glimpse sequences in each frame. These glimpses correspond to interest points in the scene that are relevant to the classified activities. No spatial coherence is forced on the glimpse locations, which gives the module liberty to explore different points at each frame and better optimize the process of scrutinizing visual information. Tracking and sequentially integrating this kind of unstructured data is a challenge, which we address by separating the set of glimpses from a set of recurrent tracking/recognition workers. These workers receive glimpses, jointly performing subsequent motion tracking and activity prediction. The glimpses are soft-assigned to the workers, optimizing coherence of the assignments in space, time and feature space using an external memory module. No hard decisions are taken, i.e. each glimpse point is assigned to all existing workers, albeit with different importance. Our methods outperform state-of-the-art methods on the largest human activity recognition dataset available to-date; NTU RGB+D Dataset, and on a smaller human action recognition dataset Northwestern-UCLA Multiview Action 3D Dataset. Our code is publicly available at https://github.com/fabienbaradel/glimpse_clouds.

Grounding referring expressions aims to locate in an image an object referred to by a natural language expression. The linguistic structure of a referring expression provides a layout of reasoning over the visual contents, and it is often crucial to align and jointly understand the image and the referring expression. In this paper, we propose a scene graph guided modular network (SGMN), which performs reasoning over a semantic graph and a scene graph with neural modules under the guidance of the linguistic structure of the expression. In particular, we model the image as a structured semantic graph, and parse the expression into a language scene graph. The language scene graph not only decodes the linguistic structure of the expression, but also has a consistent representation with the image semantic graph. In addition to exploring structured solutions to grounding referring expressions, we also propose Ref-Reasoning, a large-scale real-world dataset for structured referring expression reasoning. We automatically generate referring expressions over the scene graphs of images using diverse expression templates and functional programs. This dataset is equipped with real-world visual contents as well as semantically rich expressions with different reasoning layouts. Experimental results show that our SGMN not only significantly outperforms existing state-of-the-art algorithms on the new Ref-Reasoning dataset, but also surpasses state-of-the-art structured methods on commonly used benchmark datasets. It can also provide interpretable visual evidences of reasoning. Data and code are available at https://github.com/sibeiyang/sgmn

Automatic defect recognition is one of the research hotspots in steel production, but most of the current methods mainly extract features manually and use machine learning classifiers to recognize defects, which cannot tackle the situation, where there are few data available to train and confine to a certain scene. Therefore, in this paper, a new approach is proposed which consists of part of pretrained VGG16 as a feature extractor and a new CNN neural network as a classifier to recognize the defect of steel strip surface based on the feature maps created by the feature extractor. Our method achieves an accuracy of 99.1% and 96.0% while the dataset contains 150 images each class and 10 images each class respectively, which is much better than previous methods.

Operator-theoretic analysis of nonlinear dynamical systems has attracted much attention in a variety of engineering and scientific fields, endowed with practical estimation methods using data such as dynamic mode decomposition. In this paper, we address a lifted representation of nonlinear dynamical systems with random noise based on transfer operators, and develop a novel Krylov subspace method for estimating the operators using finite data, with consideration of the unboundedness of operators. For this purpose, we first consider Perron-Frobenius operators with kernel-mean embeddings for such systems. We then extend the Arnoldi method, which is the most classical type of Kryov subspace method, so that it can be applied to the current case. Meanwhile, the Arnoldi method requires the assumption that the operator is bounded, which is not necessarily satisfied for transfer operators on nonlinear systems. We accordingly develop the shift-invert Arnoldi method for Perron-Frobenius operators to avoid this problem. Also, we describe an approach of evaluating predictive accuracy by estimated operators on the basis of the maximum mean discrepancy, which is applicable, for example, to anomaly detection in complex systems. The empirical performance of our methods is investigated using synthetic and real-world healthcare data.

Data association is important in the point cloud registration. In this work, we propose to solve the partial-to-partial registration from a new perspective, by introducing multi-level feature interactions between the source and the reference clouds at the feature extraction stage, such that the registration can be realized without the attentions or explicit mask estimation for the overlapping detection as adopted previously. Specifically, we present FINet, a feature interaction-based structure with the capability to enable and strengthen the information associating between the inputs at multiple stages. To achieve this, we first split the features into two components, one for rotation and one for translation, based on the fact that they belong to different solution spaces, yielding a dual branches structure. Second, we insert several interaction modules at the feature extractor for the data association. Third, we propose a transformation sensitivity loss to obtain rotation-attentive and translation-attentive features. Experiments demonstrate that our method performs higher precision and robustness compared to the state-of-the-art traditional and learning-based methods. Code will be available at https://github.com/HaoXu-Work/FINet.

3D object detection is a core component of automated driving systems. State-of-the-art methods fuse RGB imagery and LiDAR point cloud data frame-by-frame for 3D bounding box regression. However, frame-by-frame 3D object detection suffers from noise, field-of-view obstruction, and sparsity. We propose a novel Temporal Fusion Module (TFM) to use information from previous time-steps to mitigate these problems. First, a state-of-the-art frustum network extracts point cloud features from raw RGB and LiDAR point cloud data frame-by-frame. Then, our TFM module fuses these features with a recurrent neural network. As a result, 3D object detection becomes robust against single frame failures and transient occlusions. Experiments on the KITTI object tracking dataset show the efficiency of the proposed TFM, where we obtain ~6%, ~4%, and ~6% improvements on Car, Pedestrian, and Cyclist classes, respectively, compared to frame-by-frame baselines. Furthermore, ablation studies reinforce that the subject of improvement is temporal fusion and show the effects of different placements of TFM in the object detection pipeline. Our code is open-source and available at https://github.com/emecercelik/Temp-Frustum-Net.git.

Machine learning-based systems are rapidly gaining popularity and in-line with that there has been a huge research surge in the field of explainability to ensure that machine learning models are reliable, fair, and can be held liable for their decision-making process. Explainable Artificial Intelligence (XAI) methods are typically deployed to debug black-box machine learning models but in comparison to tabular, text, and image data, explainability in time series is still relatively unexplored. The aim of this study was to achieve and evaluate model agnostic explainability in a time series forecasting problem. This work focused on proving a solution for a digital consultancy company aiming to find a data-driven approach in order to understand the effect of their sales related activities on the sales deals closed. The solution involved framing the problem as a time series forecasting problem to predict the sales deals and the explainability was achieved using two novel model agnostic explainability techniques, Local explainable model-agnostic explanations (LIME) and Shapley additive explanations (SHAP) which were evaluated using human evaluation of explainability. The results clearly indicate that the explanations produced by LIME and SHAP greatly helped lay humans in understanding the predictions made by the machine learning model. The presented work can easily be extended to any time

The metro ridership prediction has always received extensive attention from governments and researchers. Recent works focus on designing complicated graph convolutional recurrent network architectures to capture spatial and temporal patterns. These works extract the information of spatial dimension well, but the limitation of temporal dimension still exists. We extended Neural ODE algorithms to the graph network and proposed the STR-GODEs network, which can effectively learn spatial, temporal, and ridership correlations without the limitation of dividing data into equal-sized intervals on the timeline. While learning the spatial relations and the temporal correlations, we modify the GODE-RNN cell to obtain the ridership feature and hidden states. Ridership information and its hidden states are added to the GODESolve to reduce the error accumulation caused by long time series in prediction. Extensive experiments on two large-scale datasets demonstrate the efficacy and robustness of our model.

Strong theoretical guarantees of robustness can be given for ensembles of classifiers generated by input randomization. Specifically, an $\ell_2$ bounded adversary cannot alter the ensemble prediction generated by an additive isotropic Gaussian noise, where the radius for the adversary depends on both the variance of the distribution as well as the ensemble margin at the point of interest. We build on and considerably expand this work across broad classes of distributions. In particular, we offer adversarial robustness guarantees and associated algorithms for the discrete case where the adversary is $\ell_0$ bounded. Moreover, we exemplify how the guarantees can be tightened with specific assumptions about the function class of the classifier such as a decision tree. We empirically illustrate these results with and without functional restrictions across image and molecule datasets.

Adoption of deep neural networks in fields such as economics or finance has been constrained by the lack of interpretability of model outcomes. This paper proposes a generative neural network architecture - the parameter encoder neural network (PENN) - capable of estimating local posterior distributions for the parameters of a regression model. The parameters fully explain predictions in terms of the inputs and permit visualization, interpretation and inference in the presence of complex heterogeneous effects and feature dependencies. The use of Bayesian inference techniques offers an intuitive mechanism to regularize local parameter estimates towards a stable solution, and to reduce noise-fitting in settings of limited data availability. The proposed neural network is particularly well-suited to applications in economics and finance, where parameter inference plays an important role. An application to an asset pricing problem demonstrates how the PENN can be used to explore nonlinear risk dynamics in financial markets, and to compare empirical nonlinear effects to behavior posited by financial theory.

As more data are produced each day, and faster, data stream mining is growing in importance, making clear the need for algorithms able to fast process these data. Data stream mining algorithms are meant to be solutions to extract knowledge online, specially tailored from continuous data problem. Many of the current algorithms for data stream mining have high processing and memory costs. Often, the higher the predictive performance, the higher these costs. To increase predictive performance without largely increasing memory and time costs, this paper introduces a novel algorithm, named Online Local Boosting (OLBoost), which can be combined into online decision tree algorithms to improve their predictive performance without modifying the structure of the induced decision trees. For such, OLBoost applies a boosting to small separate regions of the instances space. Experimental results presented in this paper show that by using OLBoost the online learning decision tree algorithms can significantly improve their predictive performance. Additionally, it can make smaller trees perform as good or better than larger trees.

We introduce a method to stabilize Generative Adversarial Networks (GANs) by defining the generator objective with respect to an unrolled optimization of the discriminator. This allows training to be adjusted between using the optimal discriminator in the generator's objective, which is ideal but infeasible in practice, and using the current value of the discriminator, which is often unstable and leads to poor solutions. We show how this technique solves the common problem of mode collapse, stabilizes training of GANs with complex recurrent generators, and increases diversity and coverage of the data distribution by the generator.

A human does not have to see all elephants to recognize an animal as an elephant. On contrast, current state-of-the-art deep learning approaches heavily depend on the variety of training samples and the capacity of the network. In practice, the size of network is always limited and it is impossible to access all the data samples. Under this circumstance, deep learning models are extremely fragile to human-imperceivable adversarial examples, which impose threats to all safety critical systems. Inspired by the association and attention mechanisms of the human brain, we propose reverse adversarial examples method that can greatly improve models' robustness on unseen data. Experiments show that our reverse adversarial method can improve accuracy on average 19.02% on ResNet18, MobileNet, and VGG16 on unseen data transformation. Besides, the proposed method is also applicable to compressed models and shows potential to compensate the robustness drop brought by model quantization - an absolute 30.78% accuracy improvement.

We present a filter pruning approach for deep model compression, using a multitask network. Our approach is based on learning a a pruner network to prune a pre-trained target network. The pruner is essentially a multitask deep neural network with binary outputs that help identify the filters from each layer of the original network that do not have any significant contribution to the model and can therefore be pruned. The pruner network has the same architecture as the original network except that it has a multitask/multi-output last layer containing binary-valued outputs (one per filter), which indicate which filters have to be pruned. The pruner's goal is to minimize the number of filters from the original network by assigning zero weights to the corresponding output feature-maps. In contrast to most of the existing methods, instead of relying on iterative pruning, our approach can prune the network (original network) in one go and, moreover, does not require specifying the degree of pruning for each layer (and can learn it instead). The compressed model produced by our approach is generic and does not need any special hardware/software support. Moreover, augmenting with other methods such as knowledge distillation, quantization, and connection pruning can increase the degree of compression for the proposed approach. We show the efficacy of our proposed approach for classification and object detection tasks.

Time series are an interesting frontier for kernel-based methods, for the simple reason that there is no kernel designed to represent them and their unique characteristics in full generality. Existing sequential kernels ignore the time indices, with many assuming that the series must be regularly-spaced; some such kernels are not even psd. In this manuscript, we show that a "series kernel" that is general enough to represent irregularly-spaced multivariate time series may be built out of well-known "vector kernels". We also show that all series kernels constructed using our methodology are psd, and are thus widely applicable. We demonstrate this point by formulating a Gaussian process-based strategy - with our series kernel at its heart - to make predictions about test series when given a training set. We validate the strategy experimentally by estimating its generalisation error on multiple datasets and comparing it to relevant baselines. We also demonstrate that our series kernel may be used for the more traditional setting of time series classification, where its performance is broadly in line with alternative methods.

Cloud vendors are increasingly offering machine learning services as part of their platform and services portfolios. These services enable the deployment of machine learning models on the cloud that are offered on a pay-per-query basis to application developers and end users. However recent work has shown that the hosted models are susceptible to extraction attacks. Adversaries may launch queries to steal the model and compromise future query payments or privacy of the training data. In this work, we present a cloud-based extraction monitor that can quantify the extraction status of models by observing the query and response streams of both individual and colluding adversarial users. We present a novel technique that uses information gain to measure the model learning rate by users with increasing number of queries. Additionally, we present an alternate technique that maintains intelligent query summaries to measure the learning rate relative to the coverage of the input feature space in the presence of collusion. Both these approaches have low computational overhead and can easily be offered as services to model owners to warn them of possible extraction attacks from adversaries. We present performance results for these approaches for decision tree models deployed on BigML MLaaS platform, using open source datasets and different adversarial attack strategies.

In this paper, we treat the image generation task using an autoencoder, a representative latent model. Unlike many studies regularizing the latent variable's distribution by assuming a manually specified prior, we approach the image generation task using an autoencoder by directly estimating the latent distribution. To this end, we introduce 'latent density estimator' which captures latent distribution explicitly and propose its structure. Through experiments, we show that our generative model generates images with the improved visual quality compared to previous autoencoder-based generative models.

Even though Deep Neural Networks (DNNs) are widely celebrated for their practical performance, they possess many intriguing properties related to depth that are difficult to explain both theoretically and intuitively. Understanding how weights in deep networks coordinate together across layers to form useful learners has proven challenging, in part because the repeated composition of nonlinearities has proved intractable. This paper presents a reparameterization of DNNs as a linear function of a feature map that is locally independent of the weights. This feature map transforms depth-dependencies into simple tensor products and maps each input to a discrete subset of the feature space. Then, using a max-margin assumption, the paper develops a sample compression representation of the neural network in terms of the discrete activation state of neurons induced by s ``support vectors". The paper shows that the number of support vectors s relates with learning guarantees for neural networks through sample compression bounds, yielding a sample complexity of O(ns/epsilon) for networks with n neurons. Finally, the number of support vectors s is found to monotonically increase with width and label noise but decrease with depth.

Facial image super-resolution (SR) is an important preprocessing for facial image analysis, face recognition, and image-based 3D face reconstruction. Recent convolutional neural network (CNN) based method has shown excellent performance by learning mapping relation using pairs of low-resolution (LR) and high-resolution (HR) facial images. However, since the HR facial image reconstruction using CNN is conventionally aimed to increase the PSNR and SSIM metrics, the reconstructed HR image might not be realistic even with high scores. An adversarial framework is proposed in this study to reconstruct the HR facial image by simultaneously generating an HR image with and without blur. First, the spatial resolution of the LR facial image is increased by eight times using a five-layer CNN. Then, the encoder extracts the features of the up-scaled image. These features are finally sent to two branches (decoders) to generate an HR facial image with and without blur. In addition, local and global discriminators are combined to focus on the reconstruction of HR facial structures. Experiment results show that the proposed algorithm generates a realistic HR facial image. Furthermore, the proposed method can generate a variety of different facial images.

Point clouds are unstructured and unordered data, as opposed to images. Thus, most machine learning approach developed for image cannot be directly transferred to point clouds. In this paper, we propose a generalization of discrete convolutional neural networks (CNNs) in order to deal with point clouds by replacing discrete kernels by continuous ones. This formulation is simple, allows arbitrary point cloud sizes and can easily be used for designing neural networks similarly to 2D CNNs. We present experimental results with various architectures, highlighting the flexibility of the proposed approach. We obtain competitive results compared to the state-of-the-art on shape classification, part segmentation and semantic segmentation for large-scale point clouds.

Graph Convolutional Networks (GCNs) have shown significant improvements in semi-supervised learning on graph-structured data. Concurrently, unsupervised learning of graph embeddings has benefited from the information contained in random walks. In this paper, we propose a model: Network of GCNs (N-GCN), 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 improves state-of-the-art baselines 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.

Deep generative models of 3D shapes have received a great deal of research interest. Yet, almost all of them generate discrete shape representations, such as voxels, point clouds, and polygon meshes. We present the first 3D generative model for a drastically different shape representation --- describing a shape as a sequence of computer-aided design (CAD) operations. Unlike meshes and point clouds, CAD models encode the user creation process of 3D shapes, widely used in numerous industrial and engineering design tasks. However, the sequential and irregular structure of CAD operations poses significant challenges for existing 3D generative models. Drawing an analogy between CAD operations and natural language, we propose a CAD generative network based on the Transformer. We demonstrate the performance of our model for both shape autoencoding and random shape generation. To train our network, we create a new CAD dataset consisting of 178,238 models and their CAD construction sequences. We have made this dataset publicly available to promote future research on this topic.

When encountering novel objects, humans are able to infer a wide range of physical properties such as mass, friction and deformability by interacting with them in a goal driven way. This process of active interaction is in the same spirit as a scientist performing experiments to discover hidden facts. Recent advances in artificial intelligence have yielded machines that can achieve superhuman performance in Go, Atari, natural language processing, and complex control problems; however, it is not clear that these systems can rival the scientific intuition of even a young child. In this work we introduce a basic set of tasks that require agents to estimate properties such as mass and cohesion of objects in an interactive simulated environment where they can manipulate the objects and observe the consequences. We found that state of art deep reinforcement learning methods can learn to perform the experiments necessary to discover such hidden properties. By systematically manipulating the problem difficulty and the cost incurred by the agent for performing experiments, we found that agents learn different strategies that balance the cost of gathering information against the cost of making mistakes in different situations.

Distributional reinforcement learning (DRL) is a recent reinforcement learning framework whose success has been supported by various empirical studies. It relies on the key idea of replacing the expected return with the return distribution, which captures the intrinsic randomness of the long term rewards. Most of the existing literature on DRL focuses on problems with discrete action space and value based methods. In this work, motivated by applications in robotics with continuous action space control settings, we propose sample-based distributional policy gradient (SDPG) algorithm. It models the return distribution using samples via a reparameterization technique widely used in generative modeling and inference. We compare SDPG with the state-of-art policy gradient method in DRL, distributed distributional deterministic policy gradients (D4PG), which has demonstrated state-of-art performance. We apply SDPG and D4PG to multiple OpenAI Gym environments and observe that our algorithm shows better sample efficiency as well as higher reward for most tasks.

Micro-expression recognition (\textbf{MER}) has attracted lots of researchers' attention in a decade. However, occlusion will occur for MER in real-world scenarios. This paper deeply investigates an interesting but unexplored challenging issue in MER, \ie, occlusion MER. First, to research MER under real-world occlusion, synthetic occluded micro-expression databases are created by using various mask for the community. Second, to suppress the influence of occlusion, a \underline{R}egion-inspired \underline{R}elation \underline{R}easoning \underline{N}etwork (\textbf{RRRN}) is proposed to model relations between various facial regions. RRRN consists of a backbone network, the Region-Inspired (\textbf{RI}) module and Relation Reasoning (\textbf{RR}) module. More specifically, the backbone network aims at extracting feature representations from different facial regions, RI module computing an adaptive weight from the region itself based on attention mechanism with respect to the unobstructedness and importance for suppressing the influence of occlusion, and RR module exploiting the progressive interactions among these regions by performing graph convolutions. Experiments are conducted on handout-database evaluation and composite database evaluation tasks of MEGC 2018 protocol. Experimental results show that RRRN can significantly explore the importance of facial regions and capture the cooperative complementary relationship of facial regions for MER. The results also demonstrate RRRN outperforms the state-of-the-art approaches, especially on occlusion, and RRRN acts more robust to occlusion.

Conventional object detection models inevitably encounter a performance drop as the domain disparity exists. Unsupervised domain adaptive object detection is proposed recently to reduce the disparity between domains, where the source domain is label-rich while the target domain is label-agnostic. The existing models follow a parameter shared siamese structure for adversarial domain alignment, which, however, easily leads to the collapse and out-of-control risk of the source domain and brings negative impact to feature adaption. The main reason is that the labeling unfairness (asymmetry) between source and target makes the parameter sharing mechanism unable to adapt. Therefore, in order to avoid the source domain collapse risk caused by parameter sharing, we propose an asymmetric tri-way Faster-RCNN (ATF) for domain adaptive object detection. Our ATF model has two distinct merits: 1) A ancillary net supervised by source label is deployed to learn ancillary target features and simultaneously preserve the discrimination of source domain, which enhances the structural discrimination (object classification vs. bounding box regression) of domain alignment. 2) The asymmetric structure consisting of a chief net and an independent ancillary net essentially overcomes the parameter sharing aroused source risk collapse. The adaption safety of the proposed ATF detector is guaranteed. Extensive experiments on a number of datasets, including Cityscapes, Foggy-cityscapes, KITTI, Sim10k, Pascal VOC, Clipart and Watercolor, demonstrate the SOTA performance of our method.

This paper addresses the problem of 3D human pose estimation in the wild. A significant challenge is the lack of training data, i.e., 2D images of humans annotated with 3D poses. Such data is necessary to train state-of-the-art CNN architectures. Here, we propose a solution to generate a large set of photorealistic synthetic images of humans with 3D pose annotations. We introduce an image-based synthesis engine that artificially augments a dataset of real images with 2D human pose annotations using 3D Motion Capture (MoCap) data. Given a candidate 3D pose our algorithm selects for each joint an image whose 2D pose locally matches the projected 3D pose. The selected images are then combined to generate a new synthetic image by stitching local image patches in a kinematically constrained manner. The resulting images are used to train an end-to-end CNN for full-body 3D pose estimation. We cluster the training data into a large number of pose classes and tackle pose estimation as a K-way classification problem. Such an approach is viable only with large training sets such as ours. Our method outperforms the state of the art in terms of 3D pose estimation in controlled environments (Human3.6M) and shows promising results for in-the-wild images (LSP). This demonstrates that CNNs trained on artificial images generalize well to real images.

Modern computer vision algorithms typically require expensive data acquisition and accurate manual labeling. In this work, we instead leverage the recent progress in computer graphics to generate fully labeled, dynamic, and photo-realistic proxy virtual worlds. We propose an efficient real-to-virtual world cloning method, and validate our approach by building and publicly releasing a new video dataset, called Virtual KITTI (see http://www.xrce.xerox.com/Research-Development/Computer-Vision/Proxy-Virtual-Worlds), automatically labeled with accurate ground truth for object detection, tracking, scene and instance segmentation, depth, and optical flow. We provide quantitative experimental evidence suggesting that (i) modern deep learning algorithms pre-trained on real data behave similarly in real and virtual worlds, and (ii) pre-training on virtual data improves performance. As the gap between real and virtual worlds is small, virtual worlds enable measuring the impact of various weather and imaging conditions on recognition performance, all other things being equal. We show these factors may affect drastically otherwise high-performing deep models for tracking.

Sampling from distributions of implicitly defined shapes enables analysis of various energy functionals used for image segmentation. Recent work describes a computationally efficient Metropolis-Hastings method for accomplishing this task. Here, we extend that framework so that samples are accepted at every iteration of the sampler, achieving an order of magnitude speed up in convergence. Additionally, we show how to incorporate topological constraints.

Current image transformation and recoloring algorithms try to introduce artistic effects in the photographed images, based on user input of target image(s) or selection of pre-designed filters. These manipulations, although intended to enhance the impact of an image on the viewer, do not include the option of image transformation by specifying the affect information. In this paper we present an automatic image-transformation method that transforms the source image such that it can induce an emotional affect on the viewer, as desired by the user. Our proposed novel image emotion transfer algorithm does not require a user-specified target image. The proposed algorithm uses features extracted from top layers of deep convolutional neural network and the user-specified emotion distribution to select multiple target images from an image database for color transformation, such that the resultant image has desired emotional impact. Our method can handle more diverse set of photographs than the previous methods. We conducted a detailed user study showing the effectiveness of our proposed method. A discussion and reasoning of failure cases has also been provided, indicating inherent limitation of color-transfer based methods in the use of emotion assignment. Project Page: http://im.itu.edu.pk/affective-image-transfer/

Deep neural networks are often ignorant about what they do not know and overconfident when they make uninformed predictions. Some recent approaches quantify classification uncertainty directly by training the model to output high uncertainty for the data samples close to class boundaries or from the outside of the training distribution. These approaches use an auxiliary data set during training to represent out-of-distribution samples. However, selection or creation of such an auxiliary data set is non-trivial, especially for high dimensional data such as images. In this work we develop a novel neural network model that is able to express both aleatoric and epistemic uncertainty to distinguish decision boundary and out-of-distribution regions of the feature space. To this end, variational autoencoders and generative adversarial networks are incorporated to automatically generate out-of-distribution exemplars for training. Through extensive analysis, we demonstrate that the proposed approach provides better estimates of uncertainty for in- and out-of-distribution samples, and adversarial examples on well-known data sets against state-of-the-art approaches including recent Bayesian approaches for neural networks and anomaly detection methods.

To improve efficiency and reduce failures in autonomous vehicles, research has focused on developing robust and safe learning methods that take into account disturbances in the environment. Existing literature in robust reinforcement learning poses the learning problem as a two player game between the autonomous system and disturbances. This paper examines two different algorithms to solve the game, Robust Adversarial Reinforcement Learning and Neural Fictitious Self Play, and compares performance on an autonomous driving scenario. We extend the game formulation to a semi-competitive setting and demonstrate that the resulting adversary better captures meaningful disturbances that lead to better overall performance. The resulting robust policy exhibits improved driving efficiency while effectively reducing collision rates compared to baseline control policies produced by traditional reinforcement learning methods.

Biofilm is a formation of microbial material on tooth substrata. Several methods to quantify dental biofilm coverage have recently been reported in the literature, but at best they provide a semi-automated approach to quantification with significant input from a human grader that comes with the graders bias of what are foreground, background, biofilm, and tooth. Additionally, human assessment indices limit the resolution of the quantification scale; most commercial scales use five levels of quantification for biofilm coverage (0%, 25%, 50%, 75%, and 100%). On the other hand, current state-of-the-art techniques in automatic plaque quantification fail to make their way into practical applications owing to their inability to incorporate human input to handle misclassifications. This paper proposes a new interactive method for biofilm quantification in Quantitative light-induced fluorescence (QLF) images of canine teeth that is independent of the perceptual bias of the grader. The method partitions a QLF image into segments of uniform texture and intensity called superpixels; every superpixel is statistically modeled as a realization of a single 2D Gaussian Markov random field (GMRF) whose parameters are estimated; the superpixel is then assigned to one of three classes (background, biofilm, tooth substratum) based on the training set of data. The quantification results show a high degree of consistency and precision. At the same time, the proposed method gives pathologists full control to post-process the automatic quantification by flipping misclassified superpixels to a different state (background, tooth, biofilm) with a single click, providing greater usability than simply marking the boundaries of biofilm and tooth as done by current state-of-the-art methods.

A central mechanism in machine learning is to identify, store, and recognize patterns. How to learn, access, and retrieve such patterns is crucial in Hopfield networks and the more recent transformer architectures. We show that the attention mechanism of transformer architectures is actually the update rule of modern Hopfield networks that can store exponentially many patterns. We exploit this high storage capacity of modern Hopfield networks to solve a challenging multiple instance learning (MIL) problem in computational biology: immune repertoire classification. Accurate and interpretable machine learning methods solving this problem could pave the way towards new vaccines and therapies, which is currently a very relevant research topic intensified by the COVID-19 crisis. Immune repertoire classification based on the vast number of immunosequences of an individual is a MIL problem with an unprecedentedly massive number of instances, two orders of magnitude larger than currently considered problems, and with an extremely low witness rate. In this work, we present our novel method DeepRC that integrates transformer-like attention, or equivalently modern Hopfield networks, into deep learning architectures for massive MIL such as immune repertoire classification. We demonstrate that DeepRC outperforms all other methods with respect to predictive performance on large-scale experiments, including simulated and real-world virus infection data, and enables the extraction of sequence motifs that are connected to a given disease class. Source code and datasets: https://github.com/ml-jku/DeepRC

In applications of supervised learning applied to medical image segmentation, the need for large amounts of labeled data typically goes unquestioned. In particular, in the case of brain anatomy segmentation, hundreds or thousands of weakly-labeled volumes are often used as training data. In this paper, we first observe that for many brain structures, a small number of training examples, (n=9), weakly labeled using Freesurfer 6.0, plus simple data augmentation, suffice as training data to achieve high performance, achieving an overall mean Dice coefficient of $0.84 \pm 0.12$ compared to Freesurfer over 28 brain structures in T1-weighted images of $\approx 4000$ 9-10 year-olds from the Adolescent Brain Cognitive Development study. We then examine two varieties of heteroscedastic network as a method for improving classification results. An existing proposal by Kendall and Gal, which uses Monte-Carlo inference to learn to predict the variance of each prediction, yields an overall mean Dice of $0.85 \pm 0.14$ and showed statistically significant improvements over 25 brain structures. Meanwhile a novel heteroscedastic network which directly learns the probability that an example has been mislabeled yielded an overall mean Dice of $0.87 \pm 0.11$ and showed statistically significant improvements over all but one of the brain structures considered. The loss function associated to this network can be interpreted as performing a form of learned label smoothing, where labels are only smoothed where they are judged to be uncertain.

Most existing re-identification methods focus on learning robust and discriminative features with deep convolution networks. However, many of them consider content similarity separately and fail to utilize the context information of the query and gallery sets, e.g. probe-gallery and gallery-gallery relations, thus hard samples may not be well solved due to the limited or even misleading information. In this paper, we present a novel Context-Aware Graph Convolution Network (CAGCN), where the probe-gallery relations are encoded into the graph nodes and the graph edge connections are well controlled by the gallery-gallery relations. In this way, hard samples can be addressed with the context information flows among other easy samples during the graph reasoning. Specifically, we adopt an effective hard gallery sampler to obtain high recall for positive samples while keeping a reasonable graph size, which can also weaken the imbalanced problem in training process with low computation complexity.Experiments show that the proposed method achieves state-of-the-art performance on both person and vehicle re-identification datasets in a plug and play fashion with limited overhead.

The field of automatic video generation has received a boost thanks to the recent Generative Adversarial Networks (GANs). However, most existing methods cannot control the contents of the generated video using a text caption, losing their usefulness to a large extent. This particularly affects human videos due to their great variety of actions and appearances. This paper presents Conditional Flow and Texture GAN (CFT-GAN), a GAN-based video generation method from action-appearance captions. We propose a novel way of generating video by encoding a caption (e.g., "a man in blue jeans is playing golf") in a two-stage generation pipeline. Our CFT-GAN uses such caption to generate an optical flow (action) and a texture (appearance) for each frame. As a result, the output video reflects the content specified in the caption in a plausible way. Moreover, to train our method, we constructed a new dataset for human video generation with captions. We evaluated the proposed method qualitatively and quantitatively via an ablation study and a user study. The results demonstrate that CFT-GAN is able to successfully generate videos containing the action and appearances indicated in the captions.

Learning feature representation from discriminative local regions plays a key role in fine-grained visual classification. Employing attention mechanisms to extract part features has become a trend. However, there are two major limitations in these methods: First, they often focus on the most salient part while neglecting other inconspicuous but distinguishable parts. Second, they treat different part features in isolation while neglecting their relationships. To handle these limitations, we propose to locate multiple different distinguishable parts and explore their relationships in an explicit way. In this pursuit, we introduce two lightweight modules that can be easily plugged into existing convolutional neural networks. On one hand, we introduce a feature boosting and suppression module that boosts the most salient part of feature maps to obtain a part-specific representation and suppresses it to force the following network to mine other potential parts. On the other hand, we introduce a feature diversification module that learns semantically complementary information from the correlated part-specific representations. Our method does not need bounding boxes/part annotations and can be trained end-to-end. Extensive experimental results show that our method achieves state-of-the-art performances on several benchmark fine-grained datasets. Source code is available at https://github.com/chaomaer/FBSD.

The mixture of Gaussian distributions, a soft version of k-means , is considered a state-of-the-art clustering algorithm. It is widely used in computer vision for selecting classes, e.g., color, texture, and shapes. In this algorithm, each class is described by a Gaussian distribution, defined by its mean and covariance. The data is described by a weighted sum of these Gaussian distributions. We propose a new method, inspired by quantum interference in physics. Instead of modeling each class distribution directly, we model a class wave function such that its magnitude square is the class Gaussian distribution. We then mix the class wave functions to create the mixture wave function. The final mixture distribution is then the magnitude square of the mixture wave function. As a result, we observe the quantum class interference phenomena, not present in the Gaussian mixture model. We show that the quantum method outperforms the Gaussian mixture method in every aspect of the estimations. It provides more accurate estimations of all distribution parameters, with much less fluctuations, and it is also more robust to data deformations from the Gaussian assumptions. We illustrate our method for color segmentation as an example application.