Speaker: Hongfu Liu Brandeis University
Time: 2024-05-14 15:00-2024-05-14 16:00
Venue: FIT 1-222

Abstract:

Data-centric learning focuses on enhancing algorithmic performance from the perspective of the training data. In contrast to model-centric learning, which designs novel algorithms or optimization techniques for performance improvement with fixed training data, data-centric learning operates with a fixed learning algorithm while modifying the training data through trimming, augmenting, or other methods aligned with improving utility. Data-centric learning holds significant potential in many areas such as model interpretation, subset training set selection, data generation, noisy label detection, active learning, and others. In this talk, I will introduce our recent advances in data-centric learning based on influence functions. 

Short Bio:

Dr. Hongfu Liu is an Assistant Professor of Computer Science at Brandeis University. His research interests lie in core machine learning and AI-assisted applications. He has published over 100 papers (e.g., NeurIPS, ICLR, ICML, IJCAI, AAAI, KDD, ICDM, SDM, CIKM, CVPR, ICCV, TPAMI, and TKDE). These publications have received over 3,500 citations with an h-index of 34 according to Google Scholar as of March 2024. He has also won several awards including the First Place Award in MS-Celel-1M Grand Challenge in ICCV 2017, the NVIDIA CCS Best Student Paper Award in FG 2021, the 2021 INNS Aharon Katzir Young Investigator Award, the top reviewer in UAI 2022, the highlighted/notable Area Chair in ICLR/NeurIPS 2022/2023, and the 2022 Global Top-25 Chinese Young Scholars in AI (Data Mining Area) by Baidu Scholar. He has served as an Associate Editor of IEEE CIM and as a (Senior) Area Chair of ICLR, ICML, and NeurIPS.