Xiaoxuan Bai

About

Xiaoxuan Bai has made pioneering contributions at the intersection of deep reinforcement learning (DRL) and adversarial machine learning, a critical area for ensuring the security and robustness of AI systems. Her research focuses on understanding and constructing adversarial examples that can deceive DRL models, particularly in pathfinding and control tasks. In her highly cited work, "Adversarial Examples Construction Towards White-Box Q Table Variation in DQN Pathfinding Training" (2018, 25 citations), Bai demonstrated how subtle perturbations to inputs can manipulate Q-value estimations in Deep Q-Networks, exposing vulnerabilities in DRL-based navigation. This foundational study has become a key reference for researchers exploring the security of autonomous systems, from robot control to computer vision. By revealing how adversarial attacks can degrade decision-making in pathfinding, Bai’s work underscores the urgent need for robust training methods in real-world AI applications. Her research not only advances theoretical understanding of adversarial robustness but also provides practical insights for safeguarding DRL systems against malicious exploitation, making her a notable voice in the growing field of trustworthy AI.

Research Focus

Key Achievements

Top Papers

1. 1
   Adversarial Examples Construction Towards White-Box Q Table Variation in DQN Pathfinding Training

   25 citations · 2018

Key Collaborators

Contact & Links