Diffusion Generative Modeling: Progress and Next Steps

Location

Calvin Lab auditorium

Date

Monday, Aug. 3 – Friday, Aug. 7, 2026

Back to calendar

9 – 9:45 a.m.

Coffee and Check-in

9:45 – 10 a.m.

Welcome

10 – 10:45 a.m.

Stochastic dynamics as a proof technique

Andrea Montanari (Stanford University)

10:45 – 11:15 a.m.

Break

11:15 a.m. – 12 p.m.

High-Accuracy Sampling for Diffusion Models and Log-Concave Distributions

Fan Chen (MIT)

12 – 2 p.m.

Lunch (on your own)

2 – 2:45 p.m.

Efficient Reinforcement Learning for Diffusion Models

Yongxin Chen (Georgia Institute of Technology)

2:45 – 3:30 p.m.

Distributional Diffusion Models

Valentin de Bortoli (Google DeepMind)

3:30 – 4:15 p.m.

Finite-particles rates for drifting models

Krishna Balasubramanian (UC Davis)

4:15 – 5 p.m.

Reception

8:30 – 9 a.m.

Coffee and Check-in

9 – 9:45 a.m.

Talk by

Kirill Neklyudov (University of Montreal)

9:45 – 10:30 a.m.

Normalizing Flow Maps

Joey Bose (Imperial College London)

10:30 – 11 a.m.

Break

11 – 11:45 a.m.

Optimal Inference Schedules for Masked Diffusion Models

Jerry Li (University of Washington)

11:45 a.m. – 2 p.m.

Lunch (on your own)

2 – 2:45 p.m.

Understanding and enhancing diffusion model: a quantification of its generalizability, and a provably scalable test-time scaling method

Molei Tao (Georgia Tech)

2:45 – 3:30 p.m.

Break

3:30 – 4:15 p.m.

Global selection, local completion: a probabilistic anatomy of diffusion U-Net

Zahra Kadkhodaie (MIT)

4:15 – 5 p.m.

A new class of algorithms for trajectory inference

Aram-Alexandre Pooladian (Yale University)

8:30 – 9 a.m.

Coffee and Check-in

9 – 9:45 a.m.

Signed Rectified Flow: Negativity-Controlled Generation

Qiang Liu (UT Austin)

9:45 – 10:30 a.m.

How to Build a Modern Diffusion Language Model

Volodymyr Kuleshov (Cornell University)

10:30 – 11 a.m.

Break

11 – 11:45 a.m.

Rare event analysis via stochastic optimal control

Carles Domingo-Enrich (Microsoft Research)

11:45 a.m. – 2 p.m.

Lunch (on your own)

2 – 5 p.m.

Open Discussion

8:30 – 9 a.m.

Coffee and Check-in

9 – 9:45 a.m.

Windowed thinning and query complexity for the bouncy particle and Zigzag samplers

Jianfeng Lu (Duke University)

9:45 – 10:30 a.m.

Inference-Time Algorithms: A Theoretical Lens on Tractability and Error Propagation

Andrej Risteski (Carnegie Mellon University)

10:30 – 11 a.m.

Break

11 – 11:45 a.m.

Sampling from the Sherrington-Kirkpatrick model up to β<1/2

Holden Lee (Johns Hopkins University)

11:45 a.m. – 2 p.m.

Lunch (on your own)

2 – 2:45 p.m.

What Really Separates Autoregression and Diffusion? A Synthesis and Path Beyond

Jiaxin Shi (Meta)

2:45 – 3:30 p.m.

Break

3:30 – 4:15 p.m.

How abundant are good interpolators?

Ahmed El Alaoui (Cornell)

4:15 – 5 p.m.

Score-Based Generative Modeling without Diffusion: Langevin MCMC All the Way

Saeed Saremi (Genentech)

8:30 – 9 a.m.

Coffee and Check-in

9 – 9:45 a.m.

Diffusion in RL and robotics: how expressive policies changed how we use continuous actions

Sergey Levine (UC Berkeley)

9:45 – 10:30 a.m.

Compositional Reasoning with Diffusion Models

Yilun Du (Harvard University)

10:30 – 11 a.m.

Break

11 – 11:45 a.m.

Do we need diffusion in robotics?

Max Simchowitz (Carnegie Mellon University)

11:45 a.m. – 2 p.m.

Lunch (on your own)

2 – 2:45 p.m.

Functional Stochastic Localization

Kevin Tian (UT Austin)

2:45 – 3:30 p.m.

Break

3:30 – 4:15 p.m.

A computational phase transition for learning-to-sample from Ising models

Thuy-Duong (June) Vuong (UC San Diego)

4:15 – 5 p.m.

Sampling from spherical spin glasses: diffusions and simulated annealing

Brice Huang (Stanford University)