MSRNE Generative Modeling & Sampling Workshop | Day 2

The goal of the Generative Modeling & Sampling Workshop was to foster dialogue across communities working at the intersection of generative modeling, sampling, stochastic control, scientific machine learning, and related areas. By bringing together perspectives from machine learning, statistics, applied mathematics, physics, chemistry, and computational biology, the workshop aimed to highlight shared mathematical and algorithmic principles, explore foundational questions and emerging applications, and create opportunities for new collaborations. It served as a forum for exchanging ideas, identifying promising directions, and accelerating progress on both fundamental and applied problems. Event page: https://www.microsoft.com/en-us/research/event/msrne-generative-modeling-sampling-workshop/ Tuesday, August 11 | Theory, Sampling & Applications 0:00 Better-than-i.i.d. sampling Lester Mackey 29:20 Mean-shift interacting particle systems for sampling and quantization Youssef Marzouk 59:27 Bridging stochastic flow maps and Boltzmann generators with normalizing flows Louis Grenioux 1:29:45 Unique recovery of transport maps and vector fields from finite measure-valued data Jonah Botvinick-Greenhouse 1:46:04 Learning normalized probability models via dual score matching Florentin Guth 2:04:52 Generative inversion of spectroscopic data for materials structure elucidation Daniel Schwalbe-Koda 2:33:24 Flow sampling: Learning diffusion samplers via denoising conditional processes Aaron J. Havens 3:05:38 A mean-field framework for inference-time distributional control of diffusion models Samuel Howard 3:22:55 Diffusion annealed Langevin Monte Carlo for generative modelling and sampling Paula Cordero Encinar 3:42:04 A simple recipe for CREPE: Controlling replica exchange via neural transports Saiffudin Syed 4:17:14 Closing remarks Yuanqi Du Day 1 recording: https://youtu.be/CkEbeodP9N0