Advanced Lecture Series in Pattern Recognition

题    目 (TITLE)：Enhancing Deep Learning with Structures

讲 座 人 (SPEAKER)：Assoc. Prof. Le Song (Georgia Institute of Technology)

主 持 人 (CHAIR)： Dr. Yan-Ming Zhang

时    间 (TIME)：10:30am, January 16 (Tuesday), 2018

地    点 (VENUE)：No.1 Conference Room (3rd floor), Intelligence Building

报告摘要（ABSTRACT）：

What has made deep learning models so effective? Is it the depth of the models or something else? How can we understand deep learning better and make it even more effective? In this talk, I will argue that the structure of a deep learning model and the landscape of the optimization problem are critically important for the success of such as a model.  I will present both empirical and theoretical evidence for understanding these structure aspects of deep learning, and show that following these findings we can design novel and state-of-the-art deep learning models for a diverse range of applications including face recognition, risk management for Fintech, reasoning over dynamic knowledge graphs, and algorithm design.

报告人简介（BIOGRAPHY）：

Le Song is an Associate Professor in the Department of Computational Science and Engineering, College of Computing, and an Associate Director of the Center for Machine Learning, Georgia Institute of Technology. He received his Ph.D. in Machine Learning under the supervision of Alex Smola from University of Sydney and NICTA in 2008, and then conducted his post-doctoral research with Eric Xing and Carlos Guestrin in the Department of Machine Learning, Carnegie Mellon University, between 2008 and 2011. Before he joined Georgia Institute of Technology in 2011, he was briefly a research scientist with Fernando Pereira at Google. His principal research direction is machine learning, especially kernel and deep embedding methods, and probabilistic graphical models for large scale and complex problems, arising from artificial intelligence, network analysis, computational biology and other interdisciplinary domains. He is the recipient of the NIPS’17 Machine Learning for Molecule sna Materials Workshop Best Paper Award, the Recsys’16 Deep Learning Workshop Best Paper Award, AISTATS'16 Best Student Paper Award, IPDPS'15 Best Paper Award, NSF CAREER Award’14, NIPS’13 Outstanding Paper Award, and ICML’10 Best Paper Award. He has also served as the area chair or senior program committee for many leading machine learning and AI conferences such as ICML, NIPS, AISTATS, AAAI and IJCAI, and the associate editor for JMLR and IEEE TPAMI.

宋乐是佐治亚理工大学计算科学与工程系终身副教授，机器学习中心副主任。他于2008年在Alex Smola的指导下从悉尼大学和NICTA获得机器学习博士学位。2008年至2011年间，在卡内基梅隆大学机器学习系Eric Xing和Carlos Guestrin的指导下进行了博士后研究。在2011年加入佐治亚理工学院之前，他曾是一名Google的科学家效力于Fernando Pereira 的机器学习部门。他的主要研究方向是核函数和深度学习的嵌入方法，机器学习的大规模算法和高效系统, 以及静态和动态网络分析，人工智能，社会科学，计算生物学等跨学科领域里的大规模复杂问题的建模和求解。他获得过很多机器学习方面的顶级国际奖项,包括NIPS’17机器学习与材料科学研讨会最佳论文奖，Recsys'16深度学习与推荐系统研讨会最佳论文奖，AISTATS'16最佳学生论文奖，IPDPS'15最佳论文奖，美国国家自然基金会NSF’14杰出青年奖，NIPS'13优秀论文奖和ICML'10最佳论文奖。历任ICML，NIPS，AISTATS，AAAI，IJCAI等机器学习和AI顶尖会议的领域主席，也是机器学习顶尖杂志JMLR及IEEE TPAMI的副主编。