Wasserstein GAN (WGAN) Implementation with TensorFlow/Keras

Welcome to the Wasserstein GAN repository! This GitHub project contains a detailed implementation of Wasserstein Generative Adversarial Networks (WGAN) using TensorFlow and Keras. WGAN is a variant of GANs designed to address training stability issues and mode collapse, providing a more reliable approach to generating realistic data.

Key Features

• Comprehensive Implementation: The repository includes a complete implementation of WGAN, covering the creation of both the generator and critic networks, the Wasserstein loss function, and the training loop.
• TensorFlow/Keras: The implementation leverages the popular deep learning frameworks TensorFlow and Keras, making it accessible and easy to understand for those familiar with these tools.
• MNIST Dataset: The code is accompanied by the use of the MNIST dataset for simplicity, allowing users to quickly grasp the key concepts and apply the implementation to other datasets as needed.

Getting Started

1. Clone the Repository:

   git clone https://github.com/your-username/Wasserstein-GAN.git](https://github.com/SCHEMOX/Wasserstein-Generative-Adversarial-Network-WGAN.git
   
   cd Wasserstein-Generative-Adversarial-Network-WGAN
   

2. Install Dependencies:

   pip install -r requirements.txt
   

3. Run the Jupyter Notebook: Open the provided Jupyter Notebook to explore the code, run training, and generate images.

Usage

Feel free to customize the implementation based on your specific requirements. Experiment with different hyperparameters, network architectures, and datasets to achieve optimal results for your use case.

Contributing

If you find any issues or have enhancements to suggest, we welcome contributions! Please fork the repository, make your changes, and submit a pull request.

License

This project is licensed under the Affero GNU General Public License (AGPL) - see the LICENSE file for details.

Acknowledgments

• The implementation is inspired by the Wasserstein GAN paper and the TensorFlow/Keras documentation.
• Special thanks to the contributors and open-source community for valuable insights and feedback.

Happy generating with Wasserstein GAN! If you have any questions or suggestions, feel free to open an issue.