Theme: Geometrization of AI (TDAI Speaker Series)
Workshop: Foundations of Large Language Models — Feature Learning, Representation, and Model Steering
Speaker: Misha Belkin, University of California, San Diego
Date & Time: Thursday, February 19, 2026 · 10:00 AM–12:00 PM (ET)
Location: Pomerene Hall Room 350 (Project Zone)
Food: Morning coffee, light refreshments, and lunch for registered attendees
Host: Dena Asta, Subhadeep Paul (TDAI Geometrization of AI Theme Leads)
This interactive workshop will build on the previous day’s seminar and focus on emerging research directions in the theory and practice of large language models, with particular attention to overparameterization, interpolation, and representation learning.
The session will feature short talks from Ohio State faculty and lightning presentations from graduate students working on LLM-related topics, including feature learning, internal representations, and methods for monitoring and steering model behavior. These brief presentations will set the stage for guided group discussion and collaborative exploration of how geometric and statistical perspectives can help explain generalization, robustness, and controllability in modern neural networks.
Participants will engage in structured discussion and research prompts, drawing connections between theory and practice and sharing perspectives across disciplines. The workshop is designed to foster dialogue among faculty, postdocs, and students, and to highlight ongoing and emerging work at Ohio State related to the Geometrization of AI.
About the Speaker:
Dr. Misha Belkin is a leading researcher in machine learning theory and representation learning, with foundational contributions to deep learning, optimization, and the geometry of high-dimensional models. His work focuses on understanding how modern AI systems learn, generalize, and organize information internally, with recent emphasis on feature learning and interpretability of large language models.
Learn more: http://misha.belkin-wang.org/