|
|
Xian Wu
Research Scientist
Meta AI
Email: xianwu2024 [at] u [dot] northwestern [dot] edu
|
I am currently a Research Scientist at Meta AI. I earned my Ph.D. in Computer Science from Northwestern University, where I was advised by Dr. Xinyu Xing.
Research Interests: LLM reasoning and AI Security.
Selected Publications
GPO: Learning from Critical Steps to Improve LLM Reasoning
Jiahao Yu, Zelei Cheng, Xian Wu†, Xinyu Xing († indicates corresponding author)
The 38th Conference on Neural Information Processing Systems (NeurIPS), 2025
(Featured in MIT Technology Review China)
BlockScan: Detecting Anomalies in Blockchain Transactions
Xian Wu*, Jiahao Yu*, Hao Liu, Wenbo Guo, Xinyu Xing (* indicates equal contribution)
The 38th Conference on Neural Information Processing Systems (NeurIPS), 2025
ReLeak: RL-based Red-teaming for LLM Privacy Leakage
Yuzhou Nie, Zhun Wang, Ye Yu, Xian Wu, Xuandong Zhao, Nathaniel D. Bastian, Wenbo Guo, Dawn Song
Second Conference on Language Modeling (COLM), 2025
Soft-Label Integration for Robust Toxicity Classification
Zelei Cheng, Xian Wu, Jiahao Yu, Xin-Qiang Cai, Shuo Han, Xinyu Xing
The 37th Annual Conference on Neural Information Processing Systems (NeurIPS), Vancouver, December 2024
RICE: Breaking Through the Training Bottlenecks of Reinforcement Learning with Explanation
Xian Wu*, Zelei Chen*, Jiahao Yu, Sabrina Yang, Gang Wang, Xinyu Xing
(* indicates equal contribution)
The 41st International Conference on Machine Learning (ICML), Vienna, July 2024
(Spotlight Presentation, top 3.5%; selected for OpenAI’s PaperBench)
StateMask: Explainable Deep Reinforcement Learning through State Mask
Xian Wu*, Zelei Chen*, Jiahao Yu, Wenbo Guo, Xinyu Xing
(* indicates equal contribution)
The 36th Annual Conference on Neural Information Processing Systems (NeurIPS), New Orleans, December 2023
PATROL: Provable Defense against Adversarial Policy in Two-player Games
Wenbo Guo*, Xian Wu*, Lun Wang, Dawn Song, Xinyu Xing
(* indicates equal contribution)
The 32nd USENIX Security Symposium (USENIX Security), Anaheim, CA, August 2023
From Grim Reality to Practical Solution: Malware Classification in Real-World Noise
Xian Wu, Wenbo Guo, Jia Yan, Baris Coskun, Xinyu Xing
The 44th IEEE Symposium on Security and Privacy (S&P), San Francisco, CA, May 2023
EDGE: Explaining Deep Reinforcement Learning Policies
Wenbo Guo, Xian Wu, Usmann Khan, Xinyu Xing
The 35th Annual Conference on Neural Information Processing Systems (NeurIPS), Virtual Conference, December 2021
BACKDOORL: Backdoor Attack against Competitive Reinforcement Learning
Lun Wang, Zaynah Javed, Xian Wu, Wenbo Guo, Xinyu Xing, Dawn Song
The 30th International Joint Conference on Artificial Intelligence (IJCAI), 2021
Adversarial Policy Learning in Two-player Competitive Games
Wenbo Guo, Xian Wu, Sui Huang, Xinyu Xing
Proceedings of The 38th International Conference on Machine Learning (ICML), 2021
Adversarial Policy Training against Deep Reinforcement Learning
Xian Wu*, Wenbo Guo*, Hua Wei*, Xinyu Xing
(* indicates equal contribution)
Proceedings of The 30th USENIX Security Symposium (USENIX Security), 2021
Ruling StarCraft Game Spitefully -- Exploiting the Blind Spot of AI-Powered Game Bots
Wenbo Guo, Xian Wu, Xinyu Xing, Jimmy Su
Proceedings of The 23rd BlackHat USA (BlackHat USA), 2020
Professional Servies
- Program Committee Member:
- NeurIPS'24,NeurIPS'25
- ICML 2025
- ICLR 2025,ICLR 2026
- AISTATS 2025, AISTATS 2026
- AAAI 2025
- KDD 2024
- CCS 2026
