Autonomous Learning from Knowledge Graphs

演讲简介：

Abstract: Embracing human knowledge into the learning process is of great importance and interests to understand the consequence of deep learning’s Cornucopla. In this talk, I will present our recent attempts on developing techniques for autonomous learning from knowledge graphs, which target to simultaneously learn with and discover knowledge from graph-structured data, like knowledge graphs. Specifically, I will introduce our methods on make hyper-parameter tuning efficient, model search automated and learned subgraph structure interpretable. Our methods have been published in, e.g., Pattern (Cell), TPAMI, NeurIPS, ACL. Besides, our methods are also the best solution on knowledge graph tasks of open-graph-benchmarks.

Bio: Dr. Quanming Yao currently is a tenure-track assistant professor at Department of Electronic Engineering, Tsinghua University. Before that, he spent three years from a researcher to a senior scientist in 4Paradigm INC, where he set up and led the company's machine learning research team. He obtained his Ph.D. degree at the Department of Computer Science and Engineering of Hong Kong University of Science and Technology (HKUST) and received his bachelor degree at HuaZhong University of Science and Technology (HUST). He has published 60+ top conference and journal papers, with more than 4300 citations and h-index 26. He regularly serves as area chairs for ICML, NeurIPS, ICLR and ACML. He is also a receipt of National Youth Talent Plan (China), Forbes 30 Under 30 (China), Young Scientist Awards (Hong Kong Institution of Science), and Google Fellowship (in machine learning).

嘉宾介绍：

姚权铭 清华大学 助理教授

姚权铭现任清华大学电子工程系助理教授、特别研究员和博士生导师，是“国家海外高层次人才引进计划青年项目”资助者。申请人于香港科技大学计算机系取得博士学位，导师为James T. Kwok教授（IEEE Fellow）；博士毕业之后，申请人加入第四范式担任科学家，在香港科技大学杨强教授（AAAI / ACM / IEEE Fellow）指导下，创建和领导公司的机器学习研究组，为国内早一批从事自动化机器学习的研究团队。目前为止，申请人已在国际权威会议和期刊发表CCF-A类论文47篇，包括3篇IEEE TPAMI、2篇IEEE TKDE、1篇JMLR、7篇NeurIPS、6篇KDD、5篇ICML、5篇IJCAI、3篇AAAI和1篇CVPR等；作者发表论文20篇、通信作者论文20篇；Web of Science总引用1700余次，Google Scholar总引用3400余次。其中，抗噪标签算法Co-teaching（NeurIPS 2018）为当年10大高引论文之一，小样本领域概述论文（CSUR 2020）为ESI热点论文（前0.1%高被引），AutoCross（KDD 2019）在银行自动化推荐场景中被广泛应用，自动知识图谱嵌入方法AutoSF（ICDE 2021）和PAS（CIKM 2021）为OGB（由图灵奖得主Bengio牵头的图学习方向榜单）榜单问鼎算法。另外，申请人在国际重要机器学习会议ICML、ICLR、AAAI和IJCAI上担任领域主席（Area Chair）；机器学习重要期刊Neural Network任职编委（Action Editor），IEEE TPAMI的AutoML特刊任职客座编委（Guest Editor）；NeurIPS 2018年自动化机器学习比赛组织者之一。申请人获得过的荣誉如下：Google全球博士奖研金（2016）、香港科技大学博士论文奖（2018-2019）、吴文俊人工智能学会青年奖（2019）、福布斯30Under30精英榜（中国区，2020）、香港科学会青年科学家（2020）、国家海外高层次人才引进计划青年项目（2021）。