Haibo Yang

Short Biography

Haibo Yang is an Assistant Professor in the Dept. of Computing and Information Sciences Ph.D. at Golisano College of Computing and Information Sciences in Rochester Institute of Technology. He received his Ph.D degree from the Dept. of Electrical and Computer Engineering at Ohio State University, advised by Prof. Jia(Kevin) Liu.

Immediate openings for fully funded Ph.D. students:

Openings for Ph.D. students are available and expected to be filled as early as Spring 2024 or Fall 2024. Students with strong backgrounds in statistics, mathematics, optimization, computer science, electrical engineering and physics are encouraged to apply. Please drop me an email if you are interested in working with me!

Research Interests

My research generally centers around distributed/federated learning. More specifically, I am interested in providing fundamental understandings of how distributed/federated learning algorithms are influenced by statistical variability and system variability due to privacy, communication and other constrains. Inspired by the theoretical insights, I am seeking to develop intelligent and efficient algorithms to better fit large-scale distributed/federated system.

Optimization algorithms.
Communication-Efficient Distributed/Federated Learning.
Robust Machine Learning.

News

[09/23] A paper on federated multi-objective learning is accepted by NeurIPS 2023 (acceptance rate: 26.1%).
[08/23] Joined the GCCIS at RIT as Tenure Track Assistant Professor.
[10/22] A paper on the PAC non-learnability of federated learning under partial client participation and the mitigation by using server-aided federated learning is accepted by NeurIPS Workshop on Recent Advances and New Challenges in Federated Learning (FL-NeurIPS'22).
[10/22] Glad to receive NeurIPS 2022 Scholar Award, see you in New Orleans!
[9/22] Our paper on mitigating fat-tailed noise (with potentially infinite variance) in federated learning is accepted by NeurIPS 2022 (acceptance rate 25.6%).
[9/22] Our paper on communication-efficient federated min-max learning is accepted by NeurIPS 2022 (acceptance rate 25.6%).
[7/22] Our paper on achieving linear convergence speedup for fully decentralized federated learning is accepted by ACM MobiHoc 2022 (acceptance rate: 19.8%).
[5/22] Our paper titled ``Anarchic Federated Learning" is accepted by ICML 2022 as Long presentation (long presentation rate: 2%, acceptance rate: 21.9%).

Haibo Yang

Short Biography

Research Interests

Contact Information

Rm 74-1073,

102 Lomb Memorial Dr.

Rochester, NY, U.S.A.

hbycis [at] rit.edu

585-475-2571

News