行知论坛221:Towards Automated Embedding of Knowledge Graphs

时间:2019-09-27浏览:10设置


报告题目:

Towards Automated Embedding of Knowledge Graphs


报告人:姚权铭 博士 (第四范式)

报告时间:2019109日(周三)10:20 - 11:00

报告地点:计算机学院4001

主办单位:计算机学院


摘要:

Knowledge graph (KG) embedding is a fundamental problem in mining relational patterns. It aims to encode the entities and relations in KG into low dimensional vector space that can be used for subsequent algorithms. Currently, there are two essential perspectives of KG embedding. The first one is the scoring function, which measures the plausibility of links between entities based on a given relation; and the second one is negative sampling, which selects unobserved triples from the given KG as negative samples for model training. In this talk, we will present our recent progress on making KG embedding automated from above two perspectives. We will show how automated machine learning (AutoML) techniques can be transferred and developed into KG domain. Such new methods can help us discover new task- and dataset-dependent KG embedding models, which can consistently outperform human-designed ones across different benchmarks.


讲者简介:

Dr. Quanming Yao is currently a principal researcher in 4Paradigm (Hong Kong) and managers company's machine learning research group. He obtained his Ph.D. degree in the Department of Computer Science and Engineering at Hong Kong University of Science and Technology (HKUST) in 2018, and received his bachelor degree at Huazhong University of Science and Technology (HUST) in 2013. He is the 1st runner up of Ph.D. Research Excellence Award (School of Engineering, HKUST, 2018-2019) and Google Fellowship (machine learning, 2016). He has 30 top-tier journal and conference papers, including ICML, NeurIPS, JMLR, TPAMI, KDD and CVPR. He was an outstanding reviewer of Neurocomputing in 2017; served as a program committee of many prestigious conferences, including ICML, NeurIPS, CVPR, AAAI, and IJCAI; one of the committees of AutoML competition in NeurIPS-2018, IJCNN-2019.




返回原图
/