时间：2023年5月29日 14:00-15:30

地点：电院群楼2-410室 

Learning with Noisy Labels

TongLiang Liu  

Associate Professor, Mohamed bin Zayed University

摘要:

Label noise is ubiquitous in the era of big data. Large machine learning models can easily overfit label errors, and thus cannot generalize well without properly handling the noise. In this talk, we will introduce a typical approach to deal with label noise, i.e., modelling label noise, which is essential to build statistically consistent classifiers. Specifically, given only noisy data, we will show how to model and estimate the label transition relationship from the latent clean labels to noisy labels. We hope that the participants will roughly know how to learn with noisy labels via the talk.

简介:

Tongliang Liu is currently an Associate Professor in the Department of Machine Learning at the Mohamed bin Zayed University of AI. He is also the Director of Sydney AI Centre at the University of Sydney. He is broadly interested in the fields of trustworthy machine learning and its interdisciplinary applications, with a particular emphasis on learning with noisy labels, adversarial learning, transfer learning, unsupervised learning, and statistical deep learning theory. He has authored and co-authored more than 100 research articles including ICML, NeurIPS, ICLR, CVPR, AAAI, IJCAI, JMLR, and TPAMI. His monograph on machine learning with noisy labels will be published by MIT Press. He is/was a (senior-) meta reviewer for many conferences, such as ICML, NeurIPS, ICLR, UAI, AAAI, IJCAI, and KDD, and is a notable AC for ICLR. He is an Associate Editors of TMLR and at the Editorial Boards of JMLR and MLJ. He is a recipient of AI’s 10 to Watch Award from IEEE in 2023, the Future Fellowship Award from Australian Research Council (ARC) in 2022, and the Discovery Early Career Researcher Award (DECRA) from ARC in 2018.