Conjugate Energy-Based Models

TL;DR

Conjugate energy-based models (CEBMs) introduce a new class of models that define a joint density over data and latent variables, offering flexible data similarity measures without a generator network, and show competitive results in image modeling and out-of-domain detection.

Contribution

The paper presents conjugate energy-based models (CEBMs), a novel class that combines energy-based modeling with conjugate priors, eliminating the need for generator networks and enhancing flexibility.

Findings

Achieve competitive image modeling performance

Demonstrate effective latent space predictive power

Show improved out-of-domain detection

Abstract

In this paper, we propose conjugate energy-based models (CEBMs), a new class of energy-based models that define a joint density over data and latent variables. The joint density of a CEBM decomposes into an intractable distribution over data and a tractable posterior over latent variables. CEBMs have similar use cases as variational autoencoders, in the sense that they learn an unsupervised mapping from data to latent variables. However, these models omit a generator network, which allows them to learn more flexible notions of similarity between data points. Our experiments demonstrate that conjugate EBMs achieve competitive results in terms of image modelling, predictive power of latent space, and out-of-domain detection on a variety of datasets.