October 16, 2026 at 2 PM
The SMAC team are hosting SCOOL
Abstract:
Reinforcement learning is an exciting field of research, at the crossing of Statistics, Optimization, Control and Representation theory, and is often considered the Brain of an AI, where decisions are computed and optimized. Despite this appeal, theory people often say RL is still far from applicable in practice. In this talk, I will explore how the last decade of research has shaped an unprecedented research agenda, bridging core statistical theory to applied sciences, nurturing real-life-adapted algorithms and formalisms, all within the small data/low computational budget/noisy feedback/risk-aware/high-reproducibility and high-fidelity setting. I will show how simple experimental science questions make appear exciting theory questions, and highlight the construction of an international shared effort in making RL research matter to experimental sciences at large.
Salle Agora 1, bâtiment ESPRIT
français
English