Score-based Denoising Diffusion with Non-Isotropic Gaussian Noise Models
This is an incremental improvement for researchers in generative AI, focusing on a more general noise formulation.
The paper tackles the problem of generative modeling by extending denoising diffusion techniques to use non-isotropic Gaussian noise models, instead of the standard isotropic ones, and demonstrates through experiments on CIFAR-10 that this approach can produce high-quality samples.
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 CIFAR-10 dataset to help verify empirically that this more general modeling approach can also yield high-quality samples.