Talks

Talk 1: Dr. Antonio Ortega, University of Southern California

Title: Graph Signal Processing for Machine Learning Applications: New Insights and Algorithms

Abstract:

Graph signal processing (GSP) is an active area of research that seeks to extend to signals defined on irregular graphs tools concepts such as frequency, filtering and sampling that are well understood for conventional signals defined on regular grids. As an example this leads to the definition of so called, graph Fourier transforms (GFTs). In this talk we will provide an introduction to basic GSP concepts developed over the last few years. Then we will investigate how GSP concepts can allow us to view machine learning problems from a different perspective. Specifically, we will discuss our recent work in i) graph representations that take into account local data structure ii) graph signal sampling interpretations of semi-supervised learning, and iii) a GSP-based analysis of deep learning systems.

Bio:

Antonio Ortega received his undergraduate and doctoral degrees from Universidad Politécnica de Madrid, Madrid, Spain and Columbia University, New York, NY, respectively. In 1994 he joined the Electrical and Computer Engineering department at the University of Southern California (USC), where he is currently a Professor and has served as Associate Chair. He is a Fellow of the IEEE and EURASIP, and a member of ACM and APSIPA. He is currently a member of the Board of Governors of the IEEE Signal Processing Society. He has received several paper awards, including the 2016 Signal Processing Magazine award. His recent research work is focusing on graph signal processing, machine learning, multimedia compression and wireless sensor networks.

Talk 2: Dr. Bruno Ribeiro, Purdue University

Title: Unearthing the Relationship Between Graph Neural Networks and Matrix Factorization

Abstract:

Graph tasks are ubiquitous, with applications ranging from recommendation systems, to language understanding, to automation with environmental awareness and molecular synthesis. A fundamental challenge in applying machine learning to these tasks has been encoding (representing) the graph structure in a way that ML models can easily exploit the relational information in the graph, including node and edge features. Until recently, this encoding has been performed by factor models (a.k.a. graph embeddings), which originated in 1904 with Spearman's common factors. Recently, however, graph neural networks have introduced a new powerful way to encode graphs for machine learning models. In my tutorial, I will describe these two approaches and then introduce a unifying mathematical framework that connects them. Using this novel framework, I will introduce new practical guidelines to generating and using node embeddings and graph representations, which fixes significant shortcomings of the standard operating procedures used today.

Bio:

Bruno Ribeiro is an Assistant Professor in the Department of Computer Science at Purdue University. He obtained his Ph.D. at the University of Massachusetts Amherst and did his postdoctoral studies at Carnegie Mellon University from 2013-2015. His research interests are in machine learning and data mining, with a focus on sampling and modeling relational and temporal data.