Prof. Mingsheng Long
The School of Software, Tsinghua University, Beijing, China
We address the problem of domain adaptation from theoretical and algorithmic perspectives. Existing domain adaptation theories naturally imply minimax optimization algorithms, which connect well with the domain adaptation methods based on adversarial learning. However, several disconnections still exist and form the gap between theory and algorithm. We extend previous theories to multiclass classification in domain adaptation, where classifiers based on the scoring functions and margin loss are standard choices in algorithm design. We introduce Margin Disparity Discrepancy, a novel measurement with rigorous generalization bounds, tailored to the distribution comparison with the asymmetric margin loss, and to the minimax optimization for easier training. Our theory can be seamlessly transformed into an adversarial learning algorithm for domain adaptation, successfully bridging the gap between theory and algorithm. A series of empirical studies show that our algorithm achieves the state of the art accuracies on challenging domain adaptation tasks. Short Bio: Mingsheng Long is an Associate Professor in the School of Software, Tsinghua University. He is the director of the Machine Learning Group, National Engineering Laboratory for Big Data Software. He received the B.E. degree in Electrical Engineering and the Ph.D. degree in Computer Science from Tsinghua University in 2008 and 2014 respectively. He was a visiting researcher in the AMP Lab and SAIL Lab at UC Berkeley. His research spans machine learning theories, algorithms, and systems, with special interest in deep learning, transfer learning, predictive learning, and adversarial learning. He has published 45 papers in top conferences and journals such as TPAMI, ICML, and NIPS, which have received more than 3000 citations in Google Scholar. He is serving as an area chair of ICLR, as senior PC members of IJCAI/AAAI, and as PC members of the ICML, NIPS, CVPR, etc. He received the Distinguished Dissertation Award from China Association for Artificial Intelligence in 2016 and the First Prize of the Technical Invention Award from the Ministry of Education in 2018 (the 4 th contributor).