Improved Contrastive Divergence Training of Energy Based Models
This work offers an incremental improvement to the stability and performance of training energy-based models, which could benefit researchers and practitioners working with these models.
This paper addresses training instabilities in contrastive divergence for energy-based models by re-introducing and estimating a previously omitted gradient term. The authors demonstrate improved performance on image generation, OOD detection, and compositional generation benchmarks.
Contrastive divergence is a popular method of training energy-based models, but is known to have difficulties with training stability. We propose an adaptation to improve contrastive divergence training by scrutinizing a gradient term that is difficult to calculate and is often left out for convenience. We show that this gradient term is numerically significant and in practice is important to avoid training instabilities, while being tractable to estimate. We further highlight how data augmentation and multi-scale processing can be used to improve model robustness and generation quality. Finally, we empirically evaluate stability of model architectures and show improved performance on a host of benchmarks and use cases,such as image generation, OOD detection, and compositional generation.