RBM (Restricted Boltzmann Machine) Example

This repository contains an example implementation of a Restricted Boltzmann Machine (RBM) using TensorFlow. The RBM is trained on the MNIST dataset for image reconstruction.

Overview

The RBM class is implemented with methods for training the RBM, reconstructing input data, and generating output. It utilizes Gibbs sampling with Contrastive Divergence for training. The RBM architecture consists of a visible layer and a hidden layer with binary units.

Dependencies

The code requires the following libraries:

• TensorFlow
• NumPy
• Matplotlib

Usage

Clone the repository and navigate to the project directory.

git clone https://github.com/pyritez3/DL-RBM.git

cd DL-RBM

You can install the dependencies using the following command:

pip install tensorflow numpy matplotlib

Results

The script will train the RBM for a specified number of epochs and plot the reconstruction error over epochs. Additionally, it will show a comparison of original and reconstructed images using random samples from the test data.

Customize

You can modify the hyperparameters such as learning rate, batch size, number of hidden units, and training epochs by changing the corresponding values in the RBM class instantiation (rbm = RBM(...)) and the train method call (rbm.train(...)).

Feel free to experiment with different configurations and datasets to explore the capabilities of RBMs.

License

This project is licensed under the MIT License