Paper ID: 2411.07233

Score-based generative diffusion with "active" correlated noise sources

Alexandra Lamtyugina, Agnish Kumar Behera, Aditya Nandy, Carlos Floyd, Suriyanarayanan Vaikuntanathan

Diffusion models exhibit robust generative properties by approximating the underlying distribution of a dataset and synthesizing data by sampling from the approximated distribution. In this work, we explore how the generative performance may be be modulated if noise sources with temporal correlations -- akin to those used in the field of active matter -- are used for the destruction of the data in the forward process. Our numerical and analytical experiments suggest that the corresponding reverse process may exhibit improved generative properties.

Submitted: Nov 11, 2024