Score-based Denoising Diffusion with Non-Isotropic Gaussian Noise Models

Vikram Voleti, Christopher Pal, Adam M Oberman

Keywords: Denoising diffusion probabilistic models, score-based, image generation

TL;DR: We lay the theoretical foundations of non-isotropic Gaussian noise models in score-based denoising diffusion. We use one class of non-isotropic Gaussian called Gaussian Free Fields, and show comparable results.

Abstract: Generative models based on denoising diffusion techniques have led to an unprecedented increase in the quality and diversity of imagery that is now possible to create with neural generative models. However, most contemporary state-of-the-art methods are derived from a standard isotropic Gaussian formulation. In this work we examine the situation where non-isotropic Gaussian distributions are used. We present the key mathematical derivations for creating denoising diffusion models using an underlying non-isotropic Gaussian noise model. We also provide initial experiments with the CIFAR10 dataset to help verify empirically that this more general modelling approach can also yield high-quality samples.

Student Paper: Yes