1、简介

本人因为关注图学习领域发展，于是将自己感兴趣部分进行mark。了解到LOGS邀请到了KDD2022 PhD thesis winner来自耶鲁大学助理教授应智韬，为我们带来一期双曲表示学习与知识图谱的精彩报告。

报告嘉宾：Rex Ying (应智韬@耶鲁大学）

报告题目

Hyperbolic Embeddings and Knowledge Graphs

报告摘要

The first part of the talk introduces the concept of geometric embeddings and hyperbolic embeddings. The second part covers its application in knowledge graphs.

Hierarchical relations are prevalent and indispensable for organizing human knowledge captured by a knowledge graph (KG). Here we present ConE (Cone Embedding), a KG embedding model that is able to simultaneously model multiple hierarchical as well as non-hierarchical relations in a knowledge graph. ConE embeds entities into hyperbolic cones and models relations as transformations between the cones. Experiments on standard knowledge graph benchmarks show that ConE obtains state-of-the-art performance on hierarchical reasoning tasks as well as knowledge graph completion task on hierarchical graphs.

报告人简介

Rex Ying is an assistant professor in the Department of Computer Science at Yale University. His research focus includes algorithms for graph neural networks, geometric embeddings and explainable models. He is the author of many widely used GNN algorithms such as GraphSAGE, PinSAGE and GNNExplainer. In addition, he has worked on a variety of applications of graph learning in physical simulations, social networks, knowledge graphs and biology. He developed the first billion-scale graph embedding services at Pinterest, and the graph-based anomaly detection algorithm at Amazon.

He obtained my Ph.D degree in computer science at Stanford University, advised by Jure Leskovec. His thesis focuses on expressive, scalable and explainable GNNs (graph neural networks), which is available on Github. Prior to that, he graduated from Duke University in 2016 with the highest distinction. He majored in computer science and mathematics.

斯坦福大学博士 RexYing（应智韬）的博士论文《Towards Expressive and Scalable Deep Representation Learning for Graphs》拿到了今年KDD的博士论文奖。

论文链接：https://www.proquest.com/openview/43f56ba4da9db357c08883ba102092fa/1?pq-origsite=gscholar&cbl=18750&diss=y

在这篇论文中，应智韬提出了一系列方法，这些方法率先使用图神经网络来解决图表示学习在可解释性、可伸缩性和表达性方面的挑战。论文的第一部分展示了 GraphSAGE 框架，它是一个通用但功能强大的图神经网络框架。第二部分在 GraphSAGE 框架下展示了一系列工作，通过使用层次结构、几何嵌入空间以及 multi-hop 注意力来提高 GNN 的表达能力。最后，他展示了 GNN 在推荐系统、异常检测和物理模拟领域的各种应用。

应智韬于 2022 年 1 月获得斯坦福大学计算机系博士学位，师从斯坦福大学计算机学院副教授、图神经网络大牛 Jure Leskovec（拿到过 KDD 最佳论文奖、时间检验奖等奖项）。目前，应智韬已经加入耶鲁大学担任助理教授，继续研究图神经网络、表示学习、几何深度学习等方向。

除了这篇论文之外，还有两篇论文拿到了最佳论文奖亚军，分别是清华大学博士裘捷中的《Graph Representation Learning: Spectral Theory and Self-supervised Learning》和慕尼黑理工大学博士 Daniel Zügner 的《Graph Representation Learning: Spectral Theory and Self-supervised Learning》。其中，裘捷中的导师是唐杰教授，他本人目前在腾讯担任高级研究员，主要研究方向是图数据的算法设计和表示学习。