Best Arm Identification in Generalized Linear Bandits via Hybrid Feedback
In Proceedings of the 40th Annual Conference on Neural Information Processing Systems
Profile
My background.
I am a Ph.D. candidate in Computer Science at City University of Hong Kong, advised by Prof. Jinhang Zuo. Previously, I worked as a research assistant with Prof. Sibo Wang at the Chinese University of Hong Kong from July to October 2024.
I received my bachelor’s degree in Computer Science and Technology from the School of the Gifted Young at the University of Science and Technology of China, where I was advised by Prof. Xue Chen and Prof. Jinhang Zuo.
News
Recent research updates.
-
Our paper WEREWOLF: Reputation-Aware Red-Teaming for Self-Organizing LLM Multi-Agent Systems was accepted to Findings of EMNLP 2026.
-
Our paper One Rounding Fits All: Memory-Efficient Approximation Algorithms for Partition-Constrained Influence Maximization was accepted to KDD 2026.
-
Our paper Fusing Reward and Dueling Feedback in Stochastic Bandits was accepted to ICML 2025.
Research Focus
I study sequential decision-making under limited, noisy, or hybrid feedback. My current work focuses on bandit learning and influence maximization.
- Learning Theory
- Influence Maximization
- Game Theory
- Multi-Agent Systems
Selected Publications
⋆ Equal contribution.
Practical Adversarial Attacks on Stochastic Bandits via Fake Data Injection
In Proceedings of the 40th Annual Conference on Neural Information Processing Systems
Service
Academic reviewing and service.
2026
Reviewer
ICLR and NeurIPS.
2025
Reviewer
ICML and NeurIPS.
Teaching
I have served as a teaching assistant for programming, data structures, algorithms, and mathematical analysis courses.
10-second Bandit
Can you identify the better arm before your rounds run out?
Round
0 / 10
Reward
0
Expected regret
hidden
Choose an arm to begin.
Reward trail
This is the exploration–exploitation dilemma in miniature. See how I study bandit feedback