Variational Deep Learning is a method of deep learning where we use Neural Networks to generate data, instead of drawing conclusions from it.
We have currrently implemented the following Autoencoders and Generative Adversarial Networks:
- Vanilla Autoencoder
- Denoising Autoencoder
- Sparse Autoencoder
- Contractive Autoencoder
- Variational Autoencoder
- Deep Convolutional Generative Adversarial Network
- Conditional Generative Adversarial Network
All the models have been benchmarked on the MNIST dataset.