Towards a Mechanistic Explanation of Diffusion Model Generalization

NeurIPS 2024 Workshop ATTRIB Submission51 Authors

Release Opt Out: No, I don't wish to opt out of paper release. My paper should be released.

Keywords: Diffusion Models, Generative Modelling, Generalization, Inductive Bias

Abstract:

We propose a mechanism for diffusion generalization based on local denoising operations. Through analysis of network and empirical denoisers, we identify local inductive biases in diffusion models. We demonstrate that local denoising operations can be used to approximate the optimal diffusion denoiser. Using a collection of patch-based, local empirical denoisers, we construct a denoiser which approximates the generalization behaviour of diffusion model denoisers over forward and reverse diffusion processes.

Submission Number: 51