|
Hongyang R. Zhang Office Address: 216 Mass Ave #332, Boston, MA 02115 Email Contact: ho.zhang@northeastern.edu |
I am an assistant professor at Northeastern University in the Khoury College of Computer Sciences. I received my Ph.D. from Stanford University in Computer Science, followed by a postdoc at the University of Pennsylvania in Statistics and Data Science at Wharton.
My research interests touch on a broad range of topics in machine learning and artificial intelligence, including scalable algorithms and optimization, large-scale networks, reinforcement learning, and ML systems. For more information, please check out the research overview page and my CV.
I enjoy working on technically challenging problems, while striving for broader impacts by 1) advancing the scientific frontier to benefit the society, and 2) fostering a positive environment to support the next generation of researchers and engineers. I support accessible and reproducible research.
Recent updates
New paper on a compositional kernel for measuring data influence in multimodal learning!
A new theoretical framework about multi-objective RL — We quantify objective conflicts through a new measure of objective conflict coefficients, and derive nearly-optimal regret bounds. We also design two practical algorithms (routed ensemble, trajectory attribution) for this problem. See also my talk slides at INFORMS’26.
Talk slides at Brown, CS Theory seminar. See also our paper (to appear at NeurIPS later this year)!
New paper on estimating mixtures of ranking models for multi-objective alignment.
New (revised) paper on a synthetic benchmark evaluation of algorithmic reasoning using GNNs and transformer models!
Slides at KDD’26 Agent4IR workshop.
Talk slides about our recent line of work on a Hessian view of supervised fine-tuning, task attribution, and RL.
New paper on efficient kernel methods for language model and data attribution.
A new paper using a self-normalization estimator for matrix completion from very sparse observations.
A list of older logs.