Hongyang Zhang photo 

Hongyang R. Zhang

Assistant Professor of Computer Science
Khoury College of Computer Sciences
Northeastern University

ho.zhang@northeastern.edu
177 Huntington Ave, Room 2211
[Google Scholar]  [Resume]

Hi! I am an assistant professor of computer science at Northeastern University. I am interested in questions related to data, machine learning, algorithms, and networks. I study machine learning methodologies including over-parametrized neural networks, multi-task and transfer learning, and data augmentation. I am a member of the Statstical Machine Learning group and the Theory group.

Over the past decade, there has been lots of progress developing algorithms and predictive models for understanding data. While these developments are promising, we don't fully understand when and why a learning algorithm might work in many cases. In my opinion, this is a significant barrier to applying the recent development to a broader society. During my PhD, I worked on understanding the dynamics of non-convex optimization algorithms such as SGD in over-parametrized neural networks (COLT 2018). My recent work has aimed at designing tangible algorithms for practical use, often taking a combination of theoretical and empirical perspectives. One particular problem I have been focusing on over the last few years is whether and how to learn from multiple datasets, since new datasets and large (pre-trained) models are becoming available online. To this end, we have developed a framework to think about what information transfer means in the context of multi-task learning (ICLR 2020, arXiv 2021). Meanwhile, we have designed fine-tuning algorithms to reduce over-fitting (NeurIPS 2021). From a different perspective, my work has examined the generalization effects of data augmentation (ICML 2020). In this direction, we are interested in quantifying the impact of modifying data, as opposed to modifying the algorithm.

Last but not least, I'm exploring topics related to network data analysis (KDD 2016), mobility networks (work in progress), and robust methods for machine learning.

Prospective students: I'm looking for PhD students who are strong in coding and passionate about pursuing research to join my group, starting in Fall 2022. If you are interested, please apply to Northeastern's CS PhD program and mention my name in the application. Please note that I am unable comment on your chance before viewing your application first.

For master's students who are already at Northeastern and interested in exploring ML and networks, we have projects that could use your help. Feel free to send me your CV and describe your interests. My response may be delayed though; I apologize in advance. Thanks!

Recent updates

Publications

(asterisk indicates alphabetical authorship or equal contribution)

Manuscripts

Conferences

Workshops

Group

I am fortunate to work with the following students.

PhD students

Dongyue Li (since Fall 2021)

Master's students

Haotian Ju (since Spring 2021)
Virender Singh (since Fall 2021)

You could also find other faculty who work on related topics at Northeatern in the Statstical Machine Learning group and the Theory group.

Teaching

Education

Blog posts

Misc

My Chinese name written in Chinese characters is 张泓洋.