The bayesian Variational Autoencoder has composed of two parts: Encoder and Decoder. The encoder takes an image as input and encodes it into a latent variable while the decoder is designed to decode this variable from latent space to generate the same image. The model uses ADAM optimizer to optimize the parameters of our model. MNIST dataset has been used to test and evaluate the performance of the model.
- Should work on both python 2.7+ and python 3.
- Python 3 is the preferred version.
-
Numpy
-
Note: Please refer to the Report/Document_VAE.pdf for technical details.