Avoiding Latent Variable Collapse With Generative Skip Models

TL;DR

This paper introduces generative skip models with skip connections to prevent latent variable collapse in VAEs, thereby enhancing the learning of meaningful data representations without sacrificing predictive performance.

Contribution

The paper proposes a novel generative skip model architecture that increases mutual information and reduces latent variable collapse in VAEs, supported by theoretical and empirical analysis.

Findings

Skip models increase mutual information between data and latent variables.

Skip models reduce latent variable collapse compared to standard VAEs.

Models maintain similar predictive performance while learning better representations.

Abstract

Variational autoencoders learn distributions of high-dimensional data. They model data with a deep latent-variable model and then fit the model by maximizing a lower bound of the log marginal likelihood. VAEs can capture complex distributions, but they can also suffer from an issue known as "latent variable collapse," especially if the likelihood model is powerful. Specifically, the lower bound involves an approximate posterior of the latent variables; this posterior "collapses" when it is set equal to the prior, i.e., when the approximate posterior is independent of the data. While VAEs learn good generative models, latent variable collapse prevents them from learning useful representations. In this paper, we propose a simple new way to avoid latent variable collapse by including skip connections in our generative model; these connections enforce strong links between the latent variables and the likelihood function. We study generative skip models both theoretically and empirically. Theoretically, we prove that skip models increase the mutual information between the observations and the inferred latent variables. Empirically, we study images (MNIST and Omniglot) and text (Yahoo). Compared to existing VAE architectures, we show that generative skip models maintain similar predictive performance but lead to less collapse and provide more meaningful representations of the data.