This anonymized respository contains code for the paper titled "Contrastive Variational Autoencoder Enhances Salient Features" submitted to ICML 2018.
Key code files are:
- utils.py, which contains general utility files
- vae_utils.py, which contains files that define and train a fully-connected VAE and cVAE
- celeb_utils.py, which contains files that define and train a convolutional VAE and cVAE
We also include jupyter notebooks that demonstrate how to use these algorithms in a variety of different settings