Scholar
Jiaxiang Li
Google Scholar ID: h5OWvc0AAAAJ
Meta
Citations & Impact
Citations
294
H-index
8
i10-index
7
Publications
20
Co-authors
11
list available
Publications
Resume (English only)
Academic Achievements
- - Publications: 'Problem-Parameter-Free Decentralized Nonconvex Stochastic Optimization' (Pacific Journal of Optimization), 'Joint Demonstration and Preference Learning Improves Policy Alignment with Human Feedback' (ICLR 2025), 'Riemannian Bilevel Optimization' (Journal of Machine Learning Research), 'A Riemannian ADMM' (Mathematics of Operations Research), 'Getting More Juice Out of the SFT Data: Reward Learning from Human Demonstration Improves SFT for LLM Alignment' (NeurIPS 2024), 'SLTrain: a sparse plus low-rank approach for parameter and memory efficient pretraining' (NeurIPS 2024), 'Zeroth-order Riemannian Averaging Stochastic Approximation Algorithms' (SIAM Journal on Optimization), 'Revisiting Zeroth-Order Optimization for Memory-Efficient LLM Fine-Tuning: A Benchmark' (ICML 2024)
- - Awards: INFORMS Computing Society Prize
- - Projects: NSF Grant 'Bi-Level Optimization for Hierarchical Machine Learning Problems: Models, Algorithms and Applications', Co-PI
Research Experience
- - Research Scientist, Meta
- - Postdoctoral Associate, Department of Electrical and Computer Engineering, University of Minnesota, Mentored by Prof. Mingyi Hong and Prof. Shuzhong Zhang
Education
- - Ph.D. in Applied Mathematics, UC Davis, Advised by Prof. Krishna Balasubramanian and Prof. Shiqian Ma
- - B.S. in Mathematics, Zhejiang University
Background
- - Research Interests: Applied Mathematics, Optimization Theory, Artificial Intelligence
- - Specialization: Algorithm design for large-scale nonconvex optimization, convergence theory for deterministic and stochastic minimax and bilevel optimization, distributed, decentralized and federated optimization algorithms, theoretical foundations of reinforcement learning, efficient pre-training and fine-tuning methods for LLMs
