学术讲座

报告题目：An introduction to learning with noisy labels

讲座时间：2022年3月30日（周三）14:00-16:00

腾讯会议：138-191-432

主讲人姓名：刘同亮 悉尼大学人工智能中心主任

邀请人：胡迪，中国人民大学高瓴人工智能学院准聘助理教授

主讲人简介：Tongliang Liu is the Director of Sydney AI Centre at the University of Sydney. He is also heading the Trustworthy Machine Learning Laboratory. 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, ICCV, ECCV, AAAI, IJCAI, KDD, IEEE T-PAMI, T-NNLS, and T-IP. He is/was a (senior-) meta reviewer for many conferences, such as ICML, NeurIPS, ICLR, UAI, AAAI, IJCAI, and KDD. He is a recipient of Discovery Early Career Researcher Award (DECRA) from Australian Research Council (ARC); the Cardiovascular Initiative Catalyst Award by the Cardiovascular Initiative; and was named in the Early Achievers Leaderboard of Engineering and Computer Science by The Australian in 2020.

报告摘要：  is ubiquitous in the era of big data. Deep learning algorithms can easily overfit the noise and thus cannot generalize well without properly handling the noise. In this talk, we will introduce the typical approaches to deal with label noise, i.e., extracting confident examples (whose labels are likely to be correct) and modelling label noise. The former one helps get rid of the incorrect labels while the latter one helps build statistically consistent classifiers. We will illustrate the intuitions of the state-of-the-art methods. We hope that the participants will roughly know how to learn with noisy labels via the talk.