Haoyu Liu

I am Haoyu Liu (刘浩宇), a Ph.D. candidate in the Department of Industrial Engineering and Operations Research (IEOR) at the University of California, Berkeley, where I am fortunate to be advised by Prof. Zeyu Zheng. Before coming to Berkeley, I worked as a full-time Quantitative Researcher at Optiver in Shanghai (2022-2023). I received my Bachelor of Economics from the Guanghua School of Management and B.S. in Mathematics from the School of Mathematical Sciences at Peking University in 2022.

Research Interests

My research focuses on the intersection of machine learning and operations research, with particular interests in:

• Generative AI: Diffusion Models, Large Language Models, Controllable Generative AI
• Stochastic Modeling & Simulation: Simulation, Inference Efficiency
• Economics: Economic Theory, A/B Testing & Experiment Design

Selected Publications

Conference Papers

1. Dual-Modality Multi-Stage Adversarial Safety Training: Robustifying Multimodal Web Agents Against Cross-Modal Attacks (DMAST)
   H. Liu, D. Li, L. Rutishauser, Z. Zheng
   Conference on Empirical Methods in Natural Language Processing (EMNLP), 2026 (accepted)
   Code

2. Spiral RoPE: Rotate Your Rotary Positional Embeddings in the 2D Plane
   H. Liu, S. Ren, T. Zhu, P. Wang, C. Xie, A. Yuille, Z. Zheng, F. Wang
   International Conference on Machine Learning (ICML), 2026
   arXiv · Code

3. Efficient Fine-Grained Guidance for Diffusion Model Based Symbolic Music Generation
   H. Liu†, T. Zhu†, Z. Wang, Z. Jiang, Z. Zheng (†co-first author)
   International Conference on Machine Learning (ICML), 2025
   🎵 Curious about the AI-generated music? Check out our demo page and listen to them!

4. Large Language Model Assisted Experiment Design with Generative Human-Behavior Agents
   H. Liu, Y. Tang, Z. Zhang, Z. Zheng, T. Zhu
   Winter Simulation Conference (WSC), 2024

Journal Articles

1. Learning to Simulate from Heavy-Tailed Distribution via Diffusion Model
   H. Liu†, T. Zhu†, N. Jia, J. He, Z. Zheng (†co-first author)
   Operations Research, 2026 (Articles in Advance)

2. Learning to Simulate Sequentially Generated Data via Neural Networks and Wasserstein Training
   (with Tingyu Zhu and Zeyu Zheng)
   ACM Transactions on Modeling and Computer Simulation, 33(3), 2023

3. On the Concavity of Consumption Function under Habit Formation
   (with Lun Li)
   Journal of Mathematical Economics, 106, 2023

Preprints

• IDA-Bench: Evaluating LLMs on Interactive Guided Data Analysis
• Structured Diffusion Models with Mixture of Gaussians as Prior Distribution

Education

• Ph.D. in Operations Research, University of California, Berkeley, 2023 - Present
• Bachelor of Economics, Peking University, 2018 - 2022
• B.S. in Mathematics, Peking University, 2018 - 2022

Professional Experience

• Quantitative Researcher (full-time), Optiver, Sep 2022 - Jul 2023
• Student Researcher (intern), Google, Sep 2025 - Nov 2025

Selected Awards

• National Scholarship, 2021
• Academic Innovation Award in Peking University, 2021
• Grand Prize (Top Award) in the “Challenge Cup” Competition of Peking University, 2019 & 2020
• May 4th Scholarship in Peking University, 2019 & 2020
• First Prize in National Mathematics Competition, 2019 & 2020
• IEOR First Year Faculty Fellowship, 2024