Yuwen Pu
Papers
1
Total Citations
6
H-Index
1
About
Yuwen Pu is a rising researcher at the forefront of artificial intelligence security, with a primary focus on adversarial robustness in multi-agent reinforcement learning (MARL) systems. In their highly cited 2024 work, *SUB-PLAY: Adversarial Policies against Partially Observed Multi-Agent Reinforcement Learning Systems*, Pu addresses critical security vulnerabilities in real-world MARL deployments—from drone swarm control to robotic arm collaboration. This paper, already garnering 6 citations shortly after publication, introduces novel adversarial policy attacks that exploit partial observability in multi-agent settings, revealing how seemingly robust systems can be manipulated. Pu’s contributions are particularly significant for safety-critical applications, where undetected adversarial behaviors could lead to catastrophic failures in autonomous coordination. By systematically exposing these attack surfaces, Pu is helping to lay the groundwork for more resilient and trustworthy multi-agent AI systems. Their work bridges the gap between theoretical security research and practical deployment challenges, making it essential reading for students and researchers working on safe reinforcement learning, adversarial machine learning, and multi-agent systems.
Research Focus
Key Achievements
Top Papers
- 1