I seek to develop a new generation of data-driven decision-making methods, theory, and systems, which tailor modern artificial intelligence towards addressing pressing societal challenges. To this end, my research aims at
(a) making deep reinforcement learning more efficient, both computationally and statistically, in a principled manner to empower its applications in critical domains, e.g., healthcare [1], transportation [2], and finance;
(b) scaling deep reinforcement learning to design and optimize societal-scale multi-agent systems, especially those involving cooperation and/or competition among a large population of humans and/or robots.
[1] Ethan X Fang, Zhaoran Wang, and Lan Wang. Fairness-Oriented Learning for Optimal Individualized Treatment Rules. Journal of the American Statistical Association, 2021.
[2] Jiayang Li, Jing Yu, Qianni Wang, Boyi Liu, Zhaoran Wang, and Yu Marco Nie. Differentiable Bilevel Programming for Stackelberg Congestion Games. Under Major Revision at Operations Research, 2022.
<aside> <img src="/icons/alien-pixel_gray.svg" alt="/icons/alien-pixel_gray.svg" width="40px" /> The recent breakthrough in deep reinforcement learning (RL), especially its superhuman-level performance in various board and video games, e.g., Go, Atari, Dota, and StarCraft, opens up a new avenue for controlling numerous complex and unknown systems with continuous state-action spaces through learning from data. Meanwhile, for practical purposes beyond game playing, deep RL suffers from outstanding challenges, which call for principled approaches.
</aside>
[Risk Awareness and Preference Alignment]
[Computation][Exploration][Exploitation][Causality]
[Safety and Robustness][Risk Awareness and Preference Alignment]