News & Events

Multiview Graph Learning: Algorithms and Applications

Selin Aviyente, Professor, Electrical and Computer Engineering, Michigan State University

601 Computer Studies Building

In many modern data science applications, relationships between data samples are well described with a graph structure. While many real-world data are intrinsically graph-structured, there are a large number of applications, particularly in biology and neuroscience, where the graph is not readily available and needs to be learned from a set of observations, i.e. graph signals. Most of the existing work on graph learning assumes homogeneous data defined on a single, undirected graph. However, in many settings, the data are heterogeneous and arise from multiple related graphs, referred to as a multiview graph. For example, neuroimaging data across multiple subjects can be modeled as a multiview graph in which each view corresponds to an individual brain connectome. In these settings, the views of the multiview graph are closely related to each other. Therefore, learning the topology of views jointly by incorporating the relationships among views can substantially improve performance over learning each view independently.

In this talk, I introduce a framework for multiview graph learning (mvGL) built on two complementary models. In the first, the views are assumed to be generated from a shared consensus graph through a perturbation function, capturing the common structure underlying all views. In the second, structural similarity across views is driven by the connections of common hub nodes, enabling a node-level characterization of cross-view relationships. I present results on both simulated data and real neuroimaging datasets, demonstrating the effectiveness of the proposed methods in recovering meaningful graph structure from multiview observations.

Bio:

Selin Aviyente received her B.S. degree in Electrical and Electronics engineering from Bogazici University, Istanbul in 1997; M.S. and Ph.D. degrees, both in Electrical Engineering: Systems, from the University of Michigan, Ann Arbor, in 1999 and 2002, respectively. She joined the Department of Electrical and Computer Engineering at Michigan State University in 2002, where she is currently James O. Fishbeck and Lee A. Morgan Professor. Her research focuses on statistical and nonstationary signal processing, higher-order data representations and network science with applications to biological signals. She has authored more than 150 peer-reviewed journal and conference papers. She is the recipient of a 2005 Withrow Teaching Excellence Award, a 2008 NSF CAREER Award and 2021 Withrow Excellence in Diversity Award. She has served as the chair of IEEE Signal Processing Society Bioimaging and Signal Processing Technical Committee and is currently serving on IEEE Signal Processing Society Machine Learning for Signal Processing Technical Committee, Steering Committees of IEEE SPS Data Science Initiative and IEEE BRAIN. She is the Area Editor for Special Issues for IEEE Signal Processing Magazine and Senior Area Editor for IEEE Transactions on Signal and Information Processing over Networks. She is a 2025 IEEE Signal Processing Society Distinguished Lecturer.