Invited Talk

Toward Lifelong Agents in Imperfect and Changing Environments

Prof. Masashi Sugiyama

RIKEN-AIP · The University of Tokyo

22 September 2026 | 14:00–15:00 | Hybrid (MVB 1.15 & Online)

Abstract

Artificial intelligence has achieved remarkable success by learning from clean, static datasets, yet real-world agents must operate under far more challenging conditions. They continuously receive imperfect supervision, encounter changing environments, and actively influence the data they observe through their own decisions. Building robust lifelong agents therefore requires integrating several traditionally separate research areas. This talk presents a unified view of key challenges: learning from weak and noisy supervision, adapting to distribution shifts while balancing stability and plasticity, and reinforcement learning under sparse, noisy, and non-stationary feedback. Rather than treating these as independent problems, I argue that they represent different facets of the same fundamental challenge---enabling agents to learn continuously throughout their lifetime.

Bio

Masashi Sugiyama received his Ph.D. in Computer Science from the Tokyo Institute of Technology, Japan, in 2001. After serving as an assistant and associate professor at the same institute, he became a professor at the University of Tokyo in 2014. Since 2016, he has also served as the director of the RIKEN Center for Advanced Intelligence Project. His research interests include theories and algorithms of machine learning, such as weakly supervised learning, distribution shift adaptation, and reinforcement learning. He was a keynote speaker at ICLR 2023, and is currently an ICML board member and a NeurIPS advisory board member.

Page updated

Google Sites

Report abuse