Yinglun Xu

PhD student, Computer science, University of Illinois, Urbana Champaign

Joined

May 2021

Names

Yinglun Xu (Preferred)

Emails

****@illinois.edu (Confirmed)

****@gatech.edu

Personal Links

Homepage

Google Scholar

Career & Education History

PhD student

Computer science, University of Illinois, Urbana Champaign (illinois.edu)

2021 – Present

Advisors, Relations & Conflicts

PhD Advisor

Gagandeep Singh

2021 – 2026

Expertise

Reinforcement Learning

2021 – 2025

Publications

SuperCoder: Assembly Program Superoptimization with Large Language Models
Anjiang Wei, Tarun Suresh, Huanmi Tan, Yinglun Xu, Gagandeep Singh, Ke Wang, Alex Aiken
- Submitted to ICLR 2026
- Readers: Everyone
Pessimistic Reward Modeling in RLHF against Reward Hacking
Yinglun Xu, Hangoo Kang, Yuxuan Wan, Tarun Suresh, Gagandeep Singh
- Submitted to ICLR 2026
- Readers: Everyone
Corruption Robust Thompson Sampling for Gaussian Bandits
Yinglun Xu, Zhiwei Wang, Gagandeep Singh
- AISTATS 2026 Poster
- Readers: Everyone
Learning a Pessimistic Reward in RLHF: KL Regularization is Not Necessary
Yinglun Xu, Hangoo Kang, Tarun Suresh, Yuxuan Wan, Gagandeep Singh
- ARLET
- Readers: Everyone
Improving Assembly Code Performance with Large Language Models via Reinforcement Learning
Anjiang Wei, Tarun Suresh, Huanmi Tan, Yinglun Xu, Gagandeep Singh, Ke Wang, Alex Aiken
- DL4C @ NeurIPS 2025 Poster
- Readers: Everyone
Two-Step Offline Preference-Based Reinforcement Learning on Explicitly Constrained Policies
Yinglun Xu, Tarun Suresh, Rohan Gumaste, David Zhu, Ruirui Li, Zhengyang Wang, Haoming Jiang, Xianfeng Tang, Qingyu Yin, Monica Xiao Cheng, Qi Zeng, Chao Zhang, Gagandeep Singh
- Accepted by TMLR
- Readers: Everyone
Universal Black-Box Targeted Reward Poisoning Attack Against Online Deep Reinforcement Learning
Yinglun Xu, Gagandeep Singh
- Accepted by TMLR
- Readers: Everyone
Two-Step Offline Preference-Based Reinforcement Learning with Constrained Actions
Yinglun Xu, Tarun Suresh, Rohan Gumaste, David Zhu, Ruirui Li, Zhengyang Wang, Haoming Jiang, Xianfeng Tang, Qingyu Yin, Monica Xiao Cheng, Qi Zeng, Chao Zhang, Gagandeep Singh
- ICLR 2025 Conference Withdrawn Submission
- Readers: Everyone
Robust Thompson Sampling Algorithms Against Reward Poisoning Attacks
Yinglun Xu, Zhiwei Wang, Gagandeep Singh
- Submitted to ICLR 2025
- Readers: Everyone
Universal Black-Box Reward Poisoning Attack against Offline Reinforcement Learning
Yinglun Xu, Rohan Gumaste, Gagandeep Singh
- ICLR 2025 Conference Withdrawn Submission
- Readers: Everyone

View all 14 publications