Zhiqiang Xu
Assistant Professor of Machine Learning
Research interests
Xu's research interests lie at the intersection of numerical computation, stochastic optimization, and Riemannian optimization. He is also interested in deep learning, clustering, community detection, topic modeling, etc. His recent ongoing works are about faster alternating least-squares for CCA, comprehensively tight analysis of gradient descent for PCA, accelerated inexact power methods, and Riemannian search for eigenvector computation. Email
Prior to joining MBZUAI, Xu was a senior research scientist with Baidu Research in China. Xu served as a reviewer for several academic activities of NeurIPS, ICML, ICLR, IJCAI, AAAI in various years.
He also has industrial experience in automatic optical inspection (AOI) for TFT-LCD panels and solar wafers, data analytics for airlines and insurances.
- Ph.D. in computer engineering from the Nanyang Technological University, Singapore.
- Publication Zhiqiang Xu
Li has authored or co-authored more than 20 research papers with more than 600 citations.
- Zhiqiang Xu and Ping Li. Faster Noisy Power Method. ALT 2022.
- Zhiqiang Xu and Ping Li. A Comprehensively Tight Analysis of Gradient Descent for PCA. NeurIPS 2021.
- Zhiqiang Xu and Ping Li. On the Riemannian Search for Eigenvector Computation. JMLR 2021.
- Zhiqiang Xu and Ping Li. On the Faster Alternating Least-Squares for CCA. AISTATS 2021.
- Zhiqiang Xu and Ping Li. Towards Practical Alternating Least-Squares for CCA. NeurIPS 2019.
- Zhiqiang Xu. Gradient descent meets shift-and-invert preconditioning for eigenvector computation. NeurIPS 2018.
Related
MBZUAI to play vanguard role at NeurIPS 2023
MBZUAI faculty will present 53 papers at the 37th Conference on Neural Information Processing Systems.
Read more
MBZUAI ranks top 30 globally in AI, CV, ML and NLP
In just a short, two-year time span since the university was founded, MBZUAI has been ranked 127…
Read moreContact faculty affairs
Interested in working with
our renowned faculty?
Fill out the below form and we will get back to you.