Implementation of variational autoencoder (AEVB) algorithm, using Lasagne framework, as in: [1] arXiv:1312.6114 [stat.ML] (Diederik P Kingma, Max Welling 2013)
A lot of codes is borrowed from https://github.com/Lasagne/Recipes/blob/master/examples/variational_autoencoder/variational_autoencoder.py
Two experiments are performed on MNIST (Binary) and Frey Faces (continuous) datasets.
To run the training:
- you could change the network architecture in the config.json file
- run the command
python vae_train.py mnist
orpython vae_train.py faces
After training, model parameters are stored and you could visualize the learned manifold by running
python vae_visualize.py mnist
or python vae_visualize.py faces