Preventing posterior collapse in variational autoencoders for text generation via decoder regularization

Alban Petit, Caio Filippo Corro

Keywords: variational autoencoder, posterior collapse, text generation

TL;DR: We propose a novel regularization method based on fraternal dropout to prevent posterior collapse in VAEs for text generation

Abstract: Variational autoencoders trained to minimize the reconstruction error are sensitive to the posterior collapse problem, that is the proposal posterior distribution is always equal to the prior. We propose a novel regularization method based on fraternal dropout to prevent posterior collapse. We evaluate our approach using several metrics and observe improvements in all the tested configurations.