Shiqian Ma
Associate Professor
Department of Computational Applied Math and Operations Research
Department of Electrical and Computer Engineering
Rice University
Also affiliated to:
Ken Kennedy Institute
Center for Computational Finance & Economic Systems
News:
- I am an Area Chair of ICML 2024, ICLR 2024 and AISTATS 2024, and a Senior Area Chair of NeurIPS 2024.
- Feb 2024, our paper Problem-Parameter-Free Decentralized Nonconvex Stochastic Optimization is available online.
- Feb 12, 2024, I gave a talk on Decentralized Bilevel Optimization in the Online Seminar Series Machine Learning NeEDS Mathematical Optimization. Here is the Youtube recording.
- Jan 2024, our paper Decentralized Bilevel Optimization has been accepted in Optimization Letters (special issue on Recent Advances in Bilevel Optimization and Its Applications). This is the first paper on this topic.
- Jan 2024, our paper A New Inexact Proximal Linear Algorithm with Adaptive Stopping Criteria for Robust Phase Retrieval has been accepted in IEEE Transactions on Signal Processing.
- Jan 2024, we have a new paper: AdaBB: Adaptive Barzilai-Borwein Method for Convex Optimization. This is an adaptive gradient method based on BB stepsize. It is parameter-free and line-search-free. It allows large stepsize and the average of the stepsizes is lower bounded by 1/L. Essentially this provides a convergent variant of the BB method for general smooth convex optimization.
- Nov 2023, we finished a new paper on decentralized bilevel optimization. We proposed a single-loop algorithm and proved its convergence rate without any heterogeneity assumptions. The paper is here.
- Oct 2023, I am now an Action Editor of Journal of Machine Learning Research (JMLR).
- Oct 2023, My paper Proximal Gradient Method for Nonsmooth Optimization over the Stiefel Manifold received the SIAM Review SIGEST Award! Coauthored with Shixiang Chen, Anthony Man-Cho So, and Tong Zhang. This is a great honor — every four years one paper from SIAM J. Optimization is selected for this award!
- We are hosting the 2024 INFORMS Optimization Society Conference. Click here for more details.
- Oct 2023, I am elected the Secretary/Treasurer of INFORMS Optimization Society.
- Fully funded PhD positions are available for Fall 2024 enrollment. If you are interested in joining my group, please submit your application here and mention my name in your application. I don’t have time to reply to inquiries, please don’t feel discouraged if you don’t receive a reply from me.
- I gave a plenary talk at Texas Colloquium on Distributed Learning. September 28-29, 2023
- I joined the editorial board of Journal of Optimization Theory and Applications.
- July 2023, our paper titled “A Riemannian smoothing steepest descent method for non-Lipschitz optimization on submanifolds” has been accepted in Mathematics of Operations Research. This is the first work considering algorithms for solving this type of problems. It finds important applications in robust subspace recovery, dictionary learning, and sparse PCA.
- July 2023, I gave a semi-plenary talk at the XVI International Conference on Stochastic Programming (ICSP)
- July 2023, my lab is awarded two NSF grants (one from CCF, the other from ECCS) to study new algorithms and theory for large-scale bilevel optimization, and decision making problems in multi-agent distributed systems with applications in energy, power, control and networks. These are collaborative projects with Dr. Kaiyi Ji at University at Buffalo. Thank you NSF!
- July 2023, our paper entitled “Federated Learning on Riemannian Manifolds” has been published in Applied Set-Valued Analysis and Optimization, special issue on Recent Trends in Optimization Methods and Their Applications. This is dedicated to Professor Henry Wolkowicz on the occasion of his 75th birthday.
- I am an Area Chair of ICML 2023 and NeurIPS 2023.