Bingcong Li

This pic will be updated soon. 

Ph.D. candidate
Department of Electrical and Computer Engineering
University of Minnesota, Twin Cities

Office: Room 460, Walter Library
Address: 117 Pleasant Street SE Minneapolis, MN 55455
Email: lixx5599 AT umn.edu

[Google Scholar] [GitHub] [LinkedIn]

About me

I am a Ph.D. candidate at University of Minnesota working with Signal Processing in Networking and Communications (SPiNCOM) supervised by Prof. Georgios B. Giannakis. Before that, I received BS degree in Electronic Engineering from Fudan University advised by Prof. Xin Wang.

Central to my research is the interests to design and apply optimization theories and tools to endow machine learning and networked systems with efficiency. In particular, the main thrusts include

  • Numerical efficiency – practical optimizers that converge fast;

  • System efficiency – advanced implementation that reduces memory, runtime, and energy requirement; and,

  • Statistical efficiency – new tools that handle uncertianty and robustness issues in machine learning systems.

Specifically, the list below summarizes the key techniques/topics that I have used to facilitate efficiency mentioned above.

  • optimization – (heavy ball and Nesterov's) momentum, stochastic optimization, Frank-Wolfe (aka conditional gradient), variance reduction;

  • distributed systems – distributed optimization, federated learning, large-scale algorithms; and,

  • learning with uncertanties – multi-armed (contextual) bandits, distributionally robust learning.

More about my past works can be found via [publications by year] or [publications by topic].

The past a few years have witnessed smart ideas and amazing progresses in machine learning and AI community. Advances there greatly expands my interests. My current work focuses on integrating recent results in optimization and distributed systems to large Natural Language Processing (NLP) models and Graph Neural Networks (GNNs).

Recent news