Schedule
This workshop will be held in-person at NeurIPS 2025 at the Room 20C of San Diego Convention Center, USA, on Saturday Dec 6th. The session will cover a tutorial, invited talks, contributed talks, posters, and a panel discussion. The schedule in local time zone can be found below.
| San Diego Time (PST) | Type | Title & Speakers |
|---|---|---|
| 8:00 - 8:30 | Coffee Break | |
| 8:30 - 8:40 | Opening Remarks | |
| 8:40 - 9:10 | Invited Talk |
Demystifying Diffusion Objectives: Reweighted Losses are Better Variational Bounds Jiaxin Shi |
| 9:10 - 9:25 | Sponser Talk | Basis Research Institute |
| 9:25 - 9:40 | Sponser Talk | Genesis Molecular AI |
| 9:40 - 10:10 | Invited Talk |
Iterative Improvement for Large-Scale Generative Models Chelsea Finn, Yoonho Lee |
| 10:10 - 10:30 | Coffee Break | |
| 10:30 - 11:00 | Invited Talk |
Probabilistic Inference is the Future of Foundation Models Andrew Gordon Wilson |
| 11:00 - 11:10 | Contributed Talks |
Any-Subset Autoregressive Models and Speculative Decoding Gabe Guo |
| 11:10 - 11:20 | Contributed Talks |
Ambient Diffusion Omni Giannis Daras |
| 11:20 - 11:30 | Contributed Talks |
Learning Boltzmann Generators via Constrained Mass Transport Denis Blessing |
| 11:30 - 12:30 | Poster Session #1 & Lunch Break | |
| 12:30 - 13:30 | Panel Session | Is probabilistic inference still relevant in the era of foundation models? Valentin De Bortoli, Carles Domingo-enrich, Xiang Lisa Li, Luhuan Wu |
| 13:30 - 14:00 | Invited Talk |
When Scaling Laws Fail: Test-time Scaling in Generative AI for Science Arash Vahdat |
| 14:00 - 14:30 | Invited Talk |
Video diffusion models as world models Ruiqi Gao |
| 14:30 - 14:40 | Coffee Break | |
| 14:40 - 15:10 | Invited Talk |
Skala: Advancing the Accuracy of Computational Chemistry with Deep Learning Giulia Luise |
| 15:10 - 15:20 | Contributed Talks |
Inference-time scaling of diffusion models through classical search Xiangcheng Zhang |
| 15:20 - 15:30 | Contributed Talks |
Posterior Inference in Latent Space for Scalable Constrained Black-box Optimization Kiyoung Om |
| 15:30 - 15:40 | Contributed Talks |
When rule learning breaks: Diffusion Fails to Learn Parity of Many Bits Binxu Wang |
| 15:40 - 15:45 | Awards, takeaways, and closing | |
| 15:45 - 17:00 | Poster Session #2 & Social |