FoMI: A Seminar on the Foundations of Machine Intelligence

About

Join us for the FoMI Seminar — a cross-disciplinary series exploring the theoretical foundations of machine learning and intelligence, spanning mathematics, statistics, physics, and computer science.

Organizer

Coming Up

Date Speaker Title Materials Location/Link
2026-04-29 15:00-16:00 (Beijing time) Maximilian Engel, University of Amsterdam
2026-05-14 10:00-11:00 (Beijing time) Elliot Paquette, McGill University

Past Events

Date Speaker Title Materials
2026-03-26 16:00-17:00 (Beijing time) Dmitry Yarotsky, Steklov Institute of Mathematics Corner gradient descent: provable acceleration of power-law convergence of SGD poster
2025-12-26 10:00-11:00 (Beijing time) Zihao Wang, Stanford University Phase transitions for feature learning in neural networks poster
2025-11-19 10:00-11:00 (Beijing time) Kun Chen, Institute of Theoretical Physics, CAS How LLM learn to reason--with an application in building a cross-domain scientific encyclopedia poster
2025-07-24 11:00-12:00 (Beijing time) Siddhartha Mishra, ETH Zurich AI for data-driven simulations in physics poster | slide
2025-07-16 10:00-11:00 (Beijing time) Johannes Schmidt-Hieber, University of Twente Overparametrization and the bias-variance dilemma poster | slide
2025-06-05 15:00-16:00 (Beijing time) Yiqiao Zhong, UW-Madison Can large language models solve compositional tasks? A study of out-of-distribution generalization poster | slide
2025-04-10 10:00-11:00 (Beijing time) Jeremy Cohen, Flatiron Institute How does gradient descent work? poster | slide