Data Science/Computational Social Science Seminar: Jiliang Tang

Understanding and Designing Graph Neural Networks as Graph Signal Denoising

Abstract:
Graph Neural Networks (GNNs) have shown their power in graph representation learning that have advanced various real-world applications in many domains such as biology and health care. As a result, a large number of GNNs have been developed in recent years. In this talk, I first connect numerous GNNs with a traditional optimization problem on graphs, i.e., the graph denoising problem. This connection not only provides a unified understanding of various existing GNNs but also paves a principled and innovative way to design GNNs as graph denoising problems. As an illustration, I then demonstrate how to design GNNs from this new perspective to advance real-world applications. 

Speaker bio: 

Jiliang Tang is an associate professor in the computer science and engineering department at Michigan State University. Before that, he was a research scientist in Yahoo Research. He got his PhD from Arizona State University in 2015 under Dr. Huan Liu, and earned his MS and BE from Beijing Institute of Technology in 2010 and 2008 respectively. His research interests include graph learning and trustworthy AI and their applications in education and biology. He authored the book Deep Learning on Graphs, published by the Cambridge University Press.  He was the recipient of the ICDM 2021 Tao Li Award, the 2021 Rising Star Award from the Association of Chinese Scholars in Computing, the 2020 ACM SIGKDD Rising Star Award, the 2019 NSF Career Award, and eight best paper awards (or runner-ups) including WSDM2018 and KDD2016. His dissertation won the 2015 KDD Best Dissertation runner-up and Dean's Dissertation Award. He serves as conference organizer (e.g., KDD, SIGIR, WSDM, and SDM) and journal editor (e.g., IEEE TKDE and ACM TKDD). He has published his research in highly ranked journals and top conference proceedings, which have received 20,000+ citations with an h-index 67 and extensive media coverage. 

Register to attend DS/CSS events at umsi.info/DSCSS.