Diffusion Models for Inverse Problems and Forecasting

Deep CosmoStat Days, CEA Paris-Saclay, February 2026


François Lanusse










slides at eiffl.github.io/talks/CosmoStat2026

Thank you for your attention!



Key Papers Presented:

DiEM & MMPS

Learning Diffusion Priors from Observations by Expectation Maximization

F. Rozet, G. Andry, F. Lanusse, G. Louppe

Multiscale Forecasting

Predicting partially observable dynamical systems via diffusion models with a multiscale inference scheme

R. Morel, F.P. Ramunno, J. Shen, A. Bietti, K. Cho, M. Cranmer, S. Golkar, O. Gugnin, G. Krawezik, T. Marwah, M. McCabe, L. Meyer, P. Mukhopadhyay, R. Ohana, L. Parker, H. Qu, F. Rozet, K.D. Leka, F. Lanusse, D. Fouhey, S. Ho

Latent Diffusion

Lost in Latent Space: An Empirical Study of Latent Diffusion Models for Physics Emulation

F. Rozet, R. Ohana, M. McCabe, G. Louppe, F. Lanusse, S. Ho