Binomial Flows: Denoising and flow matching for discrete ordinal data @MicrosoftResearch
Binomial Flows: Denoising and flow matching for discrete ordinal data  @MicrosoftResearch
Uploaded August 2026 | Updated November 2026, 3 hours ago
Flow-based generative modeling in continuous spaces exploit Tweedie's formula to express the denoiser (learned in training) as a score function (used in sampling). In contrast, this relation has been largely missing in the discrete setting where common approaches focus on learning discrete scores and rates. In this work we close this gap for discrete non-negative ordinal data by introducing Binomial flows. Our framework provides a simple recipe for training a discrete diffusion model which simultaneously denoises, samples, and estimates exact likelihoods. We verify our methodology on synthetic examples and obtain competitive results on real-world data sets.

Speaker bio: Yair Shenfeld is an Assistant Professor of Applied Mathematics at Brown University. He currently works on the foundations and applications of generative modeling and optimal transport.

Find seminar details and upcoming talks: microsoft.com/en-us/research/event/microsoft-research-new-england-generative-modeling-sampling-seminar
Binomial Flows: Denoising and flow matching for discrete ordinal dataAgent Lightning: Microsoft Research Forum #ai #llm #agenticai #reasoning #opensource #githubHow AI will accelerate biomedical research and discoveryAbstracts: April 16, 2024Scalable emulation of protein equilibrium ensembles with BioEmuInside Microsoft Research Asia: Fundamental research powering the future of AIThe Future of the Grid:  How AI Transforms Power System PlanningHow AI is reshaping the future of healthcare and medical researchAnaximander: Interactive Orchestration and Evaluation of Geospatial Foundation ModelsWatch now: The last episode of Microsoft Research Forum of 2025The Policy Layer that Governs Africa’s Algorithmic EcosystemSecurity and Privacy in the Underserved Majority World: Practices, Challenges, and Opportunities
Microsoft Research |

Binomial Flows: Denoising and flow matching for discrete ordinal data

SHARE TO X SHARE TO REDDIT SHARE TO FACEBOOK WALLPAPER