Jingwang Li

alt text 

PhD Student, Department of Electronic & Computer Engineering
The Hong Kong University of Science & Technology
Hong Kong, China

Email: jingwang.li [at] connect.ust.hk (Preferred)
             jingwangli [at] outlook.com
Links: Google Scholar | Github

About

I am currently a PhD student at the Hong Kong University of Science & Technology, advised by Prof. Vincent Lau.
Previously, I received a BMgmt in Engineering Management from Huazhong Agricultural University and an MEng in Control Science & Engineering under the supervision of Prof. Housheng Su from Huazhong University of Science & Technology. After earning my MEng, I worked as an algorithm engineer at FABU.AI, engaging in modeling and solving optimization problems related to autonomous driving. Upon deciding to pursue a PhD, I resigned and went to the Chinese University of Hong Kong, Shenzhen as a research assistant, where I worked on accelerated decentralized optimization under the mentorship of Prof. Xiao Li. See my CV for more detailed information.

News

Research Interests

My research interests primarily revolve around the theory and applications of decentralized optimization. Currently, I am particularly interested in developing communication and computation efficient algorithms to solve a class of decentralized optimization problems with globally coupled constraints. In this context, we usually consider a networked system consisting of \(n\) agents, and the underlying network topology can be represented by an undirected or directed graph. The problem of interest can be formulated as follows:

\[ \begin{equation} \begin{aligned} \min_{x_i \in \mathbb{R}^{d_i}, y \in \mathbb{R}^p} \ & \sum_{i=1}^n \left(f_i(x_i) + g_i(x_i)\right) + h(y) \\ \text{s.t.} \ & \sum_{i=1}^{n}A_ix_i = y, \end{aligned} \end{equation} \label{original_pro} \]

where \(f_i:\mathbb{R}^{d_i} \rightarrow \mathbb{R}\), \(g_i:\mathbb{R}^{d_i} \rightarrow \mathbb{R} \cup \{+\infty\}\) and \(A_i \in \mathbb{R}^{p \times d_i}\) are private for agent \(i\), while \(h:\mathbb{R}^p \rightarrow \mathbb{R} \cup \{+\infty\}\) is public for all agents. The problem \(\eqref{original_pro}\) captures a wide range of real-world scientific and engineering problems, including decentralized vertical federated learning, decentralized optimal transport, decentralized resource allocation, and more, which is exactly why it appeals to me so much.

Publications

* indicates the corresponding author.

Preprints

  1. Jingwang Li and Housheng Su*. “Decentralized constraint-coupled optimization with inexact oracle”, arXiv:2309.06330, 2023. [arXiv]

  2. Jingwang Li and Housheng Su*. “Gradient tracking: A unified approach to smooth distributed optimization”, arXiv:2202.09804, 2022. [arXiv]

Journal Articles

  1. Jingwang Li and Housheng Su*. “NPGA: A unified algorithmic framework for decentralized constraint-coupled optimization”, IEEE Transactions on Control of Network Systems, 2024. [DOI] [arXiv]

  2. Jingwang Li, Qing An and Housheng Su*. “Proximal nested primal-dual gradient algorithms for distributed constraint-coupled composite optimization”, Applied Mathematics and Computation, 2023. [DOI]

  3. Jingwang Li and Housheng Su*. “Implicit tracking-based distributed constraint-coupled optimization”, IEEE Transactions on Control of Network Systems, 2022. [DOI] [arXiv]

Conference Proceedings

  1. Bingxi Jia*, Haidong Ren, Zheng Yang, Jingwang Li and Xiaofei He. “Cascaded motion estimation for intelligent vehicles in GNSS-denied scenes”, Chinese Control Conference, 2023. [DOI]

  2. Jingwang Li, Suoxia Miao and Housheng Su*. “Distributed primal-dual mirror dynamics for constraint-coupled optimization”, International Conference on Guidance, Navigation and Control, 2022. [DOI]

Academic Services

Reviewer for

Visitors Map

The data have been collected since Oct. 2023.