12/07/2021
By Ian Chen
Title: Low Pass Graph Signal Processing - Applications and Beyond
Location: Via Zoom
Abstract: As a key development in the emerging field of graph signal processing, the notion of graph filters has been used to define generative models for graph data which explains many examples of network dynamics. With this interpretation, classical signal processing tools such as frequency analysis have been applied with analogous interpretation to graph data, generating new insights for data science. In this talk, we present a "user guide" on a specific class of graph data, where the generating graph filters are low pass such that the filter attenuates contents in the higher graph frequencies while retaining contents in the lower frequencies. Our choice is motivated by the prevalence of low pass models in application domains such as social networks, financial markets, and power systems. We demonstrate how to leverage properties of low pass graph filters to learn graph topology; efficiently represent graph data through sampling, recover missing measurements; infer high level features of graph topology (such as community, centrality) when the data is imperfect, e.g., rank deficient or with missed observations. We also discuss a data-driven approach to (blindly) detect if the graph filter is indeed low pass.
Bio: Hoi-To Wai received his B.Eng. degree and his M.Phil. degree in electronic engineering from the Chinese University of Hong Kong (CUHK) and his Ph.D. degree in electrical engineering from Arizona State University (ASU), Tempe. He is an assistant professor in the Department of Systems Engineering and Engineering Management, CUHK, and previously held research positions at ASU; UC Davis; Telecom ParisTech; Ecole Polytechnique; and MIT. His research interests include signal processing, machine learning, and distributed optimization. His papers are mainly published on prestigious conferences such as COLT, NeurIPS, ICASSP, AAAI, ACC. His dissertation received the Dean’s Dissertation Award from ASU, and he received a Best Student Paper Award at the IEEE ICASSP.
The colloquia will also be recorded. Please feel free to ask for access to all our events.