Mar 2024 (talk): I gave a talk that summarizes our recent work on multitask learning! Here is my slide.
Jan 2024 (service): Excited to serve as an area chair of ICML’24!
Nov 2023: Excited to join the Institute for Experiential AI as a core member! Thanks to Jennifer and Usama for the kind invitation!
Oct 2023 (paper): Our report on the traffic accident analysis is now online!
Sep 2023 (paper): New paper at NeurIPS. We collect a dataset to study US traffic accidents and evaluate graph neural networks’ performance of predicting accident occurrences.
May 2023 (paper): Excited about a new paper at KDD! We provide a new method for multitask learning on graphs, with an application to supervised overlapping community detection.
April 2023 (funding): Grateful for Northeastern's TIER1 program for supporting our research on predicting road accidents using GNN.
March 2023 (paper): Excited about a newly accepted paper at TMLR! We present a new perspective to tackle the problem of negative transfer in multi-task learning.
March 2023 (talk): Visiting Yale and giving a guest lecture about “Generalization in Neural Networks: Recent Trend and Future Outlook.”
February 2023 (service): I'm orgazining an INFORMS session about “Recent Trend in Machine Learning Theory and Its Application to Operations Management.” If you are interested in giving a talk in this session, please send me an email.
February 2023 (talk): I gave a talk titled “Information Transfer in Multi-Task Learning, Data Augmentation, and Beyond” at AAAI.
January 2023 (paper): Excited about a new paper that will be presented at AISTATS’23! We prove tight generalization bounds for message-passing neural networks that scale with the spectral norm of the graph diffusion matrices.
December 2022 (paper): Excited about a recent paper that will be presented at SDM’23! We present algorithms to reduce epidemic spread on weighted graphs and time-varying graphs.
December 2022 (service): Panelist of NSF Core programs.
November 2022 (talk): Excited to receive an invitation as part of AAAI’23 New Faculty Highlights!
October 2022 (talk): Talked about transfer learning and random matrix theory at INFORMS.
September 2022 (service): Excited to be an area chair of AISTATS’23!
August 2022 (service): PC member of WSDM’23 and AAAI’23.
August 2022 (paper): presented a new algorithm about reducing epidemic spread at the epiDAMIK workshop at KDD’22.
May 2022 (paper): Excited about two recent works that will be presented at ICML’22 (thanks to the anonymous referees for feedback!):
One paper with Michael Zhang et al. about a contrastive learning approach for improving worst-group performance.
Another paper with my students about Hessian based generalization for fine-tuned models.
May 2022 (talk): gave a talk about understanding and improving generalization in multitask and transfer learning at the One World ML Seminar.
Mar 2022 (service): PC member of KDD’22, reviewer of ICML’22.
Mar 2022 (talk): presenting our recent work on fine-tuning at Northeastern's CS Theory Lunch Seminar.
Feb 2022 (preprint): a new paper that proposes generalization measures and robust algorithms for fine-tuning.
Jan 2022 (teaching): advanced machine learning in the spring semester.
Dec 2021 (service): PC member of WWW’22 (social network analysis and graph algorithms track).
Nov 2021, talk: Gave a presentation to the ACM class at SJTU and a colloquium talk at the CS department at WPI.
Oct 2021: In NeurIPS’21, Dongyue and I present regularization methods to mitigate over-fitting during fine-tuning. Thanks to the anonymous reviewers and ACs for critical feedback.
Sep 2021: Welcoming Dongyue and Virender to the group!
Aug 2021: A new manuscript that explains positive (and negative) transfer using random matrix theory!
Mar 2021: Received a Khoury seed grant for investigating robustness and bias in ML algorithms (Joint w/ Huy Nguyen).
Jan 2021: Teaching Advanced Machine Learning this semester!
Sep 2020: I'm teaching an advanced course in deep learning this semester! Feedback more than welcome.
July 2020: At COLT’20, we show an interesting dynamic of gradient descent for learning over-parametrized two-layer ReLU neural nets. We found that gradient descent first learns lower-order tensors/moments and then learns higher-order tensors/moments (Yuanzhi's video)!
May 2020: At ICML’20, we describe a formal analysis of data augmentation. Inspired by the analysis, we propose an uncertainty-based sampling scheme that achieves SoTA results (Sen's code release)!
Feb 2020: Blog posts that illustrate our recent progress on multi-task learning and data augmentation! The data augmentation blog post is part of Sharon Y. Li's expository series about exciting recent progress in this direction.
Dec 2019: Our work on multi-task learning is accepted at ICLR’20. We rigorously formulate the notion of information transfer in simple models. This leads to several insights with practical implications, which are validated on the SoTA BERT model.
May 2019: I will be an assistant professor in Computer Science at Northeastern University, Boston, starting from Fall 2020.