Yuwei Zheng

Papers

Total Citations

H-Index

About

Yuwei Zheng is an emerging researcher specializing in the security and robustness of multi-agent reinforcement learning (MARL) systems, with a particular focus on adversarial vulnerabilities in cooperative AI settings. Their work addresses a critical challenge in real-world AI deployment: understanding and defending against adversarial attacks that exploit the coordination mechanisms underlying cooperative multi-agent systems. Zheng's most recognized contribution, "Attacking Cooperative Multi-Agent Reinforcement Learning by Adversarial Minority Influence," has garnered 7 citations and investigates how a small number of adversarial agents can disproportionately disrupt cooperative behavior — a subtle yet significant threat to deployed multi-agent systems. Building on this foundation, their work on "Robust Multi-Agent Reinforcement Learning by Mutual Information Regularization" proposes principled defenses against worst-case adversarial perturbations, tackling the combinatorial complexity that arises when agents may be independently perturbed. Though early in their research career, Zheng is carving out an important niche at the intersection of adversarial machine learning and multi-agent systems. Their contributions are increasingly relevant as cooperative AI frameworks move toward safety-critical real-world applications, making their work essential reading for researchers in robust and trustworthy AI.