Optimal transport for generative diffusion modeling: bridging Schrödinger and Bass

Add to your list(s) Download to your calendar using vCal

• Huyen Pham (Ecole Polytechnique Paris)
• Wednesday 12 November 2025, 10:50-11:30
• Seminar Room 1, Newton Institute.

If you have a question about this talk, please contact nobody.

SCLW01 - Bridging Stochastic Control And Reinforcement Learning: Theories and Applications

We study the problem of generating a continuous semimartingale that interpolates prescribed initial and terminal distributions. We cast this task as an optimal transport problem on path space that bridges the Schrödinger bridge and the Bass martingale formulations, yielding a diffusion whose drift and volatility are jointly calibrated to data. We derive an analytic characterization of the optimizer—Schrödinger–Bridge–Bass (SBB) diffusion—and show that it admits a representation as a stretched Brownian motion under an explicit change of measure. Building on this structure, we develop a practical computational scheme for SBB that is efficient and scalable, providing a principled framework for generative diffusion modeling. We outline applications to image synthesis and time‑series generation, highlighting SBB as a unifying lens connecting entropic and martingale optimal transport in modern generative modeling.

This talk is part of the Isaac Newton Institute Seminar Series series.

This talk is included in these lists:

• All CMS events
• Featured lists
• INI info aggregator
• Isaac Newton Institute Seminar Series
• School of Physical Sciences
• Seminar Room 1, Newton Institute
• bld31

Note that ex-directory lists are not shown.