Bridging theory and real-world decision making
We build mathematically grounded yet practical reinforcement learning systems: agents that learn reliable, robust behavior from imperfect, heterogeneous, and large-scale data, and adapt efficiently to new objectives, rewards, and constraints.
Research focus
RL theory & optimization
Principled algorithms with provable guarantees for data-efficient, reliable decision making.
Offline & data-driven RL
Learning reliable policies from fixed, previously collected data, without costly or risky online interaction.
Safe, constrained & multi-objective RL
Optimizing behavior under safety and cost constraints and multiple competing objectives.
Imitation & learning from imperfect data
Robust behavior from demonstrations, observations, and noisy or biased supervision, including reliable off-policy evaluation.
Structure-aware & generalizable RL
Exploiting the structure of decision problems (symmetries, embodiment, skills, and multiple agents) to generalize across tasks and environments.
Decision-making foundation models
Toward generalist agents pretrained across tasks that transfer to new objectives, rewards, and environments.
News
-
May 2026One paper was accepted to RLC 2026.
-
May 2026Jongmin Lee serves as an Area Chair for CoRL 2026.
-
Mar 2026Jongmin Lee serves as an Area Chair for NeurIPS 2026.
-
Jan 2026One paper was accepted to ICLR 2026.
-
Nov 2025Jongmin Lee serves as an Area Chair for ICML 2026.
-
Sep 2025One paper was accepted to NeurIPS 2025.
-
Aug 2025Jongmin Lee serves as an Area Chair for ICLR 2026.
-
Jan 2025One paper was accepted to ICLR 2025.
-
Jan 2025Jongmin Lee joined Yonsei University as an assistant professor and started DILLab.
We welcome motivated students and collaborators. See how to apply →