Provable Posterior Sampling with Score-Based Diffusion through Tilted Transport

Explore a cutting-edge lecture on provable posterior sampling techniques using score-based diffusion and tilted transport, presented by Jiequn Han from the Simons Foundation. Delivered as part of the Fourth Symposium on Machine Learning and Dynamical Systems at the Fields Institute, this 29-minute talk delves into advanced concepts at the intersection of machine learning and statistical inference. Gain insights into innovative approaches for tackling complex probabilistic modeling challenges and enhancing the accuracy of posterior sampling methods.