This project leverages AutoEncoders in PyTorch for feature extraction and classification on the MNIST dataset, demonstrating how unsupervised learning can enhance supervised tasks.
- Implements an AutoEncoder for dimensionality reduction and feature extraction.
- Uses a neural network classifier to categorize images based on learned representations.
- Trains and evaluates on the MNIST dataset, providing insights into model performance.
- Includes data visualization of the training process and prediction results.
- Clone the GitHub repository.
- Ensure Python 3.x and PyTorch are installed.
- Install additional dependencies as listed in
requirements.txt.
The project uses the MNIST dataset, a collection of handwritten digits, to train and test the model's performance.
- Run the training script to build and train the AutoEncoder and classifier models.
- Evaluate the model using the test script, which outputs accuracy, precision, recall, and F1 score metrics.
- Visualize the training process through loss and accuracy plots, and understand the model decisions with a confusion matrix.
The README includes a section on the results obtained from training, highlighting key performance metrics and visualizations like loss curves and confusion matrices.
Contributions to the project are welcome. Follow the standard fork, branch, and pull request workflow to propose changes.
This project is released under the MIT License. See the LICENSE file for more details.
- The PyTorch team for an excellent deep learning framework.
- The MNIST dataset maintainers for providing a reliable dataset used widely in machine learning research.
For more details, please refer to the project repository.
