← Search

Sumedh Ghaisas

2 accepted papers

2020

Adversarially Robust Representations with Smooth Encoders

ICLR 2020poster

This paper studies the undesired phenomena of over-sensitivity of representations learned by deep networks to semantically-irrelevant changes in data. We identify a cause for this shortcoming in the classical Variational Auto-encoder (VAE) objective, the evidence lower bound (ELBO). We show that the…

Cited by 35SourceScholar
2020

The Autoencoding Variational Autoencoder

NeurIPS 2020spotlight

Does a Variational AutoEncoder (VAE) consistently encode typical samples generated from its decoder? This paper shows that the perhaps surprising answer to this question is `No'; a (nominally trained) VAE does not necessarily amortize inference for typical samples that it is capable of generating. W…

Cited by 0SourcePDFScholar