The Geometry of Machine Learning 2026
📅 Tuesday, 8 September 2026 → Friday, 11 September 2026 in 53 days
Hybrid Harvard meeting on geometric and mathematical structures underlying machine learning.
The Geometry of Machine Learning 2026 is a four-day conference at the Harvard Center of Mathematical Sciences and Applications (CMSA), held in Room G10, 20 Garden Street, Cambridge, Massachusetts, from 8 to 11 September 2026. It runs as a hybrid event, with both in-person and Zoom webinar registration available.
The programme examines how AI systems can reason geometrically and how geometry illuminates the way neural networks process data, extending to formal verification and energy-based optimisers alongside large language models. It is organised by Michael R. Douglas and Mike Freedman of Harvard CMSA.
Confirmed speakers include Nada Amin (Harvard), Randall Balestriero (Brown), Michael Brenner (Harvard/Google), Surya Ganguli (Stanford), Boris Hanin (Princeton), Dimitry Krotov (IAS), Gabriele Poesia (Stanford), Matthieu Wyart (Johns Hopkins) and Sean Welleck (CMU, joining via Zoom). No fee is stated on the page, so confirm registration details and any cost via the official CMSA listing; the hybrid format makes remote attendance straightforward for those unable to travel to Cambridge.
