English

VAE Approximation Error: ELBO and Exponential Families

Machine Learning 2022-04-12 v4 Computer Vision and Pattern Recognition Machine Learning

Abstract

The importance of Variational Autoencoders reaches far beyond standalone generative models -- the approach is also used for learning latent representations and can be generalized to semi-supervised learning. This requires a thorough analysis of their commonly known shortcomings: posterior collapse and approximation errors. This paper analyzes VAE approximation errors caused by the combination of the ELBO objective and encoder models from conditional exponential families, including, but not limited to, commonly used conditionally independent discrete and continuous models. We characterize subclasses of generative models consistent with these encoder families. We show that the ELBO optimizer is pulled away from the likelihood optimizer towards the consistent subset and study this effect experimentally. Importantly, this subset can not be enlarged, and the respective error cannot be decreased, by considering deeper encoder/decoder networks.

Keywords

Cite

@article{arxiv.2102.09310,
  title  = {VAE Approximation Error: ELBO and Exponential Families},
  author = {Alexander Shekhovtsov and Dmitrij Schlesinger and Boris Flach},
  journal= {arXiv preprint arXiv:2102.09310},
  year   = {2022}
}

Comments

ICLR 2022 spotlight

R2 v1 2026-06-23T23:17:07.184Z