Xiyuan Wei Ph.D. Student Department of Computer Science and Engineering Texas A&M University College Station, Texas

About Me

Hello! I am Xiyuan Wei (魏希源). I am a fourth-year Ph.D. student at Texas A&M University, advised by Prof. Tianbao Yang. I received my bachelor degree from Nanjing University of Information Science and Technology in 2022. My research interest lies in Machine Learning.

Some Links:

• Email: x{my_last_name}@tamu.edu
• GitHub
• Google Scholar
• LinkedIn

Publications

(Asterisk denotes equal contribution)

• A Geometry-Aware Efficient Algorithm for Compositional Entropic Risk Minimization [paper] [code]
  Xiyuan Wei*, Linli Zhou*, Bokun Wang, Chih-Jen Lin, Tianbao Yang
  (ICML) 43rd International Conference on Machine Learning, 2026.

• Statistical Consistency and Generalization of Contrastive Representation Learning [paper]
  Yuanfan Li, Xiyuan Wei, Tianbao Yang, Yiming Ying
  (ICML) 43rd International Conference on Machine Learning, 2026.

• Breaking the Limits of Open-Weight CLIP: An Optimization Framework for Self-supervised Fine-tuning of CLIP [paper]
  Anant Mehta, Xiyuan Wei, Xingyu Chen, Tianbao Yang
  arXiv preprint, 2026.

• NeuCLIP: Efficient Large-Scale CLIP Training with Neural Normalizer Optimization [paper] [code]
  Xiyuan Wei, Chih-Jen Lin, Tianbao Yang
  (ICLR) 40th International Conference on Learning Representations, 2026.

• Advancing Interpretability of CLIP Representations with Concept Surrogate Model. [paper]
  Nhat Hoang-Xuan, Xiyuan Wei, Wanli Xing, Tianbao Yang, My T. Thai
  (NeurIPS) 39th Conference on Neural Information Processing Systems, 2025.

• AdFair-CLIP: Adversarial Fair Contrastive Language-Image Pre-training for Chest X-Rays [paper]
  Chenlang Yi, Zizhan Xiong, Qi Qi, Xiyuan Wei, Girish Bathla, Ching-Long Lin, Bobak J. Mortazavi, Tianbao Yang
  (MICCAI) 28th International Conference on Medical Image Computing and Computer Assisted Intervention, 2025.

• Model Steering: Learning with a Reference Model Improves Generalization Bounds and Scaling Laws. [paper] [code]
  Xiyuan Wei, Ming Lin, Fanjiang Ye, Fengguang Song, Liangliang Cao, My T. Thai, Tianbao Yang.
  (ICML Spotlight) 42nd International Conference on Machine Learning, 2025.

• FastCLIP: A Suite of Optimization Techniques to Accelerate CLIP Training with Limited Resources. [paper] [code]
  Xiyuan Wei, Fanjiang Ye, Ori Yonay, Xingyu Chen, Baixi Sun, Dingwen Tao, Tianbao Yang.
  arXiv preprint, 2024.

• Stability and Generalization of Stochastic Compositional Gradient Descent Algorithms. [paper]
  Ming Yang*, Xiyuan Wei*, Tianbao Yang, Yiming Ying.
  (ICML) 41st International Conference on Machine Learning, 2024.

• An Accelerated Variance-Reduced Conditional Gradient Sliding Algorithm for First-order and Zeroth-order Optimization [paper]
  Xiyuan Wei, Bin Gu, Heng Huang
  arXiv preprint, 2021.

• Black-Box Reductions for Zeroth-Order Gradient Algorithms to Achieve Lower Query Complexity. [paper]
  Bin Gu, Xiyuan Wei, Shangqian Gao, Ziran Xiong, Cheng Deng, Heng Huang.
  (JMLR) Journal of Machine Learning Research, 2021.

(Last updated: )