11:00 am–12:00 pm ERC 401
Title: A Data-Driven Prism: Multi-View Source Separation with Diffusion Model Priors
Abstract: In astrophysics, a common challenge is to disentangle distinct, unknown sources from observations. Examples of this source separation task include deblending galaxies in a crowded field, isolating possible dark-matter emissions from standard backgrounds, and separating telluric lines from stellar spectra. Traditional analyses often rely on simplified source models that fail to accurately reproduce the data. Recent advances have shown that diffusion models can directly learn complex prior distributions from noisy, incomplete data. In this work, we show that diffusion models can solve the source separation problem without explicit assumptions about the source. Our method relies only on multiple views, or the property that different sets of observations contain different linear transformations of the unknown sources. We show that our method succeeds even when no source is individually observed and the observations are noisy, incomplete, and vary in resolution. The learned diffusion models enable us to sample from the source priors, evaluate the probability of candidate sources, and draw from the joint posterior of our sources given an observation. These results highlight the potential to build data-driven priors in the era of large-scale sky surveys.
Zoom: https://uchicago.zoom.us/j/98910526387?pwd=nob2d2IMsbCCN5DaGnhZ1vgMNht3v5.1
