----------------------------------------------------------------------------------------------------

     Department of Systems Engineering and Engineering Management

                    The Chinese University of Hong Kong

----------------------------------------------------------------------------------------------------

Date: Friday, October 16, 2026, 4:30pm to 5:30pm HKT

Venue: ERB 909, The Chinese University of Hong Kong

Title:  Learning to Rank under Strategic Corruption: A Nonatomic Approach

Speaker: Qinzhen Li, National University of Singapore

Abstract:

We study a dynamic learning-to-rank problem in which $N$ agents strategically inflate the reward signals that are used to rank them, with corruption budgets chosen endogenously. The learner observes noisy manipulated outcomes and seeks to maximize total true reward over $T$ periods. Under a general supermodular ranking-reward model, we show that efficient learning remains achievable despite strategic manipulation. Specifically, a simple Experiment-Then-Commit policy attains regret bounded in $T$, namely $\mathcal O(\min\{N\log T, N^3\})$, under a strategy profile that forms an $o(1)$-equilibrium among agents as $N$ and $T$ grow.  Our analysis proceeds in three parts. First, we show that agents' strategic interactions converge to a static nonatomic game. Second, we find a nonatomic equilibrium in this limit, where agents' manipulation intensity increases with their types. Third, we transport the limit equilibrium to the finite-agent, finite-horizon setting, which yields non-asymptotic approximate-equilibrium and regret guarantees.

Biography:

Qinzhen Li is a Ph.D. candidate at the Institute of Operations Research and Analytics at the National University of Singapore, advised by Professor Yifan Feng. Her research studies policy design and learning in environments where participants respond strategically to platform rules. Her work uses tools from game theory, information design, and online learning, with applications to digital platforms and financial markets.

Everyone is welcome to attend the talk!

SEEM-5201 Website: http://seminar.se.cuhk.edu.hk