One-Line-of-Code Data Mollification Improves Optimization of Likelihood-based Generative Models
This addresses a key limitation in likelihood-based generative models for applications like image generation, though it is incremental as it adapts an existing technique from diffusion models.
The paper tackles the lower sample quality of likelihood-based generative models compared to diffusion models by introducing data mollification, which adds Gaussian noise during optimization, resulting in large improvements in FID scores on image datasets.
Generative Models (GMs) have attracted considerable attention due to their tremendous success in various domains, such as computer vision where they are capable to generate impressive realistic-looking images. Likelihood-based GMs are attractive due to the possibility to generate new data by a single model evaluation. However, they typically achieve lower sample quality compared to state-of-the-art score-based diffusion models (DMs). This paper provides a significant step in the direction of addressing this limitation. The idea is to borrow one of the strengths of score-based DMs, which is the ability to perform accurate density estimation in low-density regions and to address manifold overfitting by means of data mollification. We connect data mollification through the addition of Gaussian noise to Gaussian homotopy, which is a well-known technique to improve optimization. Data mollification can be implemented by adding one line of code in the optimization loop, and we demonstrate that this provides a boost in generation quality of likelihood-based GMs, without computational overheads. We report results on image data sets with popular likelihood-based GMs, including variants of variational autoencoders and normalizing flows, showing large improvements in FID score.