pac-bayes-vae

This repository includes the code associated to the paper:

Statistical Guarantees for Variational Autoencoders using PAC-Bayesian Theory, (Spotlight at NeurIPS 2023)

Sokhna Diarra Mbacke, Florence Clerc, Pascal Germain

The code uses the implementation of Lipschitz-continuous neural networks provided by Anil et al., 2019. Part of the code is inspired by the implementation of Lipschitz VAEs provided by Barrett et al. 2022.

Citation:

@inproceedings{mbacke2023statistical,  
  title={Statistical Guarantees for Variational Autoencoders using PAC-Bayesian Theory},  
  author={Mbacke, Sokhna Diarra and Clerc, Florence and Germain, Pascal},  
  journal={Advances in Neural Information Processing Systems},
  year={2023},  
}