File size: 4,544 Bytes
4edd013 71a6ec7 4edd013 bc731f0 052ab91 bc731f0 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 |
---
datasets:
- mnist
- emnist
metrics:
- accuracy
library_name: keras
license: gpl-3.0
---
# Model Card for Model ID
This model is a deep neural network for classifying handwritten digits (0-9) from images.
It was a submission for a coursework assignment and is built using Keras.
## Model Details
### Model Description
This model is designed to classify handwritten digits from the MNIST dataset.
It is a basic implementation and can be a starting point for further exploration and improvement.
- **Developed by:** Paul J. Aru
- **Model type:** Convolutional Neural Network (CNN)
- **License:** GNU GPLv3
<!--### Model Sources [optional]
Provide the basic links for the model.
- **Repository:** [More Information Needed]
- **Paper [optional]:** [More Information Needed]
- **Demo [optional]:** [More Information Needed] -->
## Uses
<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
### Direct Use
This model can be used to classify handwritten digits from images.
However, it is important to note that its performance may not be optimal and can be further improved.
### Out-of-Scope Use
This model is not intended for real-world applications where high accuracy and robustness are critical.
It is for learning purposes and serves as an example for my portfolio.
## Bias, Risks, and Limitations
The model may exhibit bias depending on the training data used.
The MNIST and EMNIST dataset might contain inherent biases, and the model might learn these biases.
The model might not perform well on unseen data, especially if the handwriting styles differ significantly from the training data.
This is a basic implementation and likely has limitations in accuracy and generalizability.
It serves as a starting point for further exploration and can be improved by experimenting with different architectures, hyperparameters, and data augmentation techniques.
### Recommendations
Users should be aware of the limitations of this model and not rely on it for critical tasks.
The model can be a good foundation for further development and experimentation in deep learning for handwritten digit classification.
## How to Get Started with the Model
```
from tensorflow.keras.models import load_model
import os
model=load_model("Best_Model.h5")
```
## Training Details
### Training Data
The model is trained on the MNIST and EMNIST dataset, a standard dataset for handwritten digit classification.
### Training Procedure
<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
#### Preprocessing
The images were preprocessed using data augmentation techniques such as shifting, rotation, resizing and introducing noise.
#### Training Hyperparameters
- Epoch: 50
- Batch Size: 32
#### Speeds, Sizes, Times
## Evaluation
<!-- This section describes the evaluation protocols and provides the results. -->
### Testing Data, Factors & Metrics
#### Testing Data
The datasets used for testing include:
- the MNIST dataset
- the EMNIST dataset
- a combined dataset of MNIST and EMNIST
- an augmented combined dataset of MNIST and EMNIST
#### Factors
The factors considered in the testing process are the misclassification errors, which indicate the percentage of incorrectly classified samples in each dataset.
The metrics used to measure the performance of the models are the percentage of misclassifications for each dataset.
#### Metrics
After testing all the models, the misclassification errors for each model are plotted using a bar chart.
The range between the best and worst errors is calculated, and the model with the lowest maximum error is identified as the best model.
### Results
#### Summary
In summary, my testing approach involves evaluating the models on different datasets, considering misclassification errors as the primary metric, and comparing the performance of the models to determine the best model.
## Environmental Impact
- **Hardware Type:** Apple M2 Max
- **Hours used:** 93min (*last Model*)
## Model Card Authors
Paul J. Aru |