PhD student, Stanford University
1 paper at NeurIPS 2025
Diffusion models disentangle latent sources from multiple, noisy, incomplete observations without source-specific assumptions, yielding generative priors and posterior inference across synthetic tasks and real galaxy data.
