I am Minrui Luo, currently a Ph.D. student in the Machine Learning program at the Georgia Institute of Technology, where I am fortunate to be advised by Professor Molei Tao. I received my bachelor’s degree from the Institute for Interdisciplinary Information Sciences at Tsinghua University.

My research focuses on the mathematical foundations of modern machine learning. I am actively exploring diffusion models, non-convex optimization, and reinforcement learning. I am also interested in AI safety and alignment, as well as causal inference.

Publications & Preprints

Diffusion Models

• Zeyang Zhang, Chengwei Liang, Xingyan Chen, Meiqi Gu, Minrui Luo, Jingzhao Zhang, and Tianxing He.
  Differences in Text Generated by Diffusion and Autoregressive Language Models.
  COLM 2026 Poster.
  [arXiv]

Deep Learning Theory and Non-convex Optimization

• Minrui Luo, Weihang Xu, Xiang Gao, Maryam Fazel, and Simon Shaolei Du.
  Global Convergence of Four-Layer Matrix Factorization under Random Initialization. Preprint, 2025. [arXiv]

Causal Inference + Machine Learning

• Minrui Luo and Zhiheng Zhang.
  Causal Matrix Completion under Multiple Treatments via Mixed Synthetic Nearest Neighbors.
  ICML 2026 Poster.
  [arXiv]

Safety and Alignment

• Zhiyu Sun*, Minrui Luo*, Yu Wang, Zhili Chen, and Tianxing He.
  Reverse-Engineering Model Editing on Language Models.
  ICML 2026 Poster.
  [arXiv]

• Minrui Luo*, Fuhang Kuang*, Yu Wang, Zirui Liu, and Tianxing He.
  SC-LoRA: Balancing Efficient Fine-tuning and Knowledge Preservation via Subspace-Constrained LoRA.
  Accepted to IJCNN 2026.
  [arXiv]

• Haoming Wen, Shi Chen, Qingyu Shi, Siyuan Liu, Minrui Luo, Jingzhao Zhang, and Tianxing He.
  Defending Against Malicious Finetuning by Scaling Train-time Adversarial Attacks.
  Preprint, 2026.
  [arXiv]

* Equal contribution.

Teaching

Lecture Assistant for Probability Theory at Georgia Institute of Technology, Sep. 2026 - present.

Teaching Assistant for Natural Language Processing at Tsinghua University, Sep. 2025 - Jan. 2026.