A Diffusive Classification Loss for Learning Energy-based Generative Models

TL;DR

This paper introduces Diffusive Classification (DiffCLF), a new training objective for energy-based models that improves efficiency and effectiveness, enabling better generative and sampling tasks.

Contribution

The paper proposes DiffCLF, a supervised classification-based training method for EBMs that overcomes mode blindness and enhances model fidelity and applicability.

Findings

DiffCLF accurately estimates energies in Gaussian mixtures.

Models trained with DiffCLF perform well in composition and Boltzmann sampling tasks.

DiffCLF outperforms existing EBM training methods in efficiency and quality.

Abstract

Score-based generative models have recently achieved remarkable success. While they are usually parameterized by the score, an alternative way is to use a series of time-dependent energy-based models (EBMs), where the score is obtained from the negative input-gradient of the energy. Crucially, EBMs can be leveraged not only for generation, but also for tasks such as compositional sampling or building Boltzmann Generators via Monte Carlo methods. However, training EBMs remains challenging. Direct maximum likelihood is computationally prohibitive due to the need for nested sampling, while score matching, though efficient, suffers from mode blindness. To address these issues, we introduce the Diffusive Classification (DiffCLF) objective, a simple method that avoids blindness while remaining computationally efficient. DiffCLF reframes EBM learning as a supervised classification problem across noise levels, and can be seamlessly combined with standard score-based objectives. We validate the effectiveness of DiffCLF by comparing the estimated energies against ground truth in analytical Gaussian mixture cases, and by applying the trained models to tasks such as model composition and Boltzmann Generator sampling. Our results show that DiffCLF enables EBMs with higher fidelity and broader applicability than existing approaches.