Events Calendar

ECE Seminar: Filter Banks on Arbitrary Graphs Using Generalized Laplacian Eigenvectors

Speaker: Dr. Eduardo Pavez Carvelli, Postdoctoral Research Associate/Ming Hsieh Department of Electrical and Computer Engineering

Talk Title: Filter Banks on Arbitrary Graphs Using Generalized Laplacian Eigenvectors

Abstract: In the past decade, Graph Signal Processing (GSP) has become a popular framework to represent and process irregular and unstructured data, such as 3D point clouds arising from immersive communication and autonomous vehicle applications.
 
In this talk I will present our recent work on two channel filter banks for signals on graphs. Filter banks and other multi-resolution transformations have been extensively used for signal and image processing. Due to graph irregularities it can be challenging to construct filter banks for graphs, while satisfying desirable properties such as critical sampling, perfect reconstruction and low complexity. Bipartite filter banks are amongst the most popular designs satisfying these requirements, yet they are limited to bipartite graphs represented by their normalized Laplacian matrix. In practice graphs are rarely bipartite and other graph matrices are often preferred.
 
We substantially extend bipartite filter bank theory to arbitrary (non-bipartite) graphs and positive definite graph matrices. Our key insight is realizing the limitations of graph eigenvectors for spectral design of graph filter banks. As an alternative, we propose to use Q-orthogonal generalized eigenvectors of graphs, which can be constructed with certain spectral symmetries that can be exploited for filter bank design. The proposed graph filter banks are applied to graphs constructed on 3D point clouds with hundreds of thousands of nodes.

Biography: Eduardo Pavez Carvelli received the B.S. and M.Sc. degrees in electrical engineering from the University of Chile, Santiago, Chile, in 2011 and 2013, respectively, and the Ph.D. degree in electrical engineering from the University of Southern California, in 2019. He was an intern at Microsoft Research, and Mitsubishi Electric Research Laboratories, and he is currently a Postdoctoral Research Associate at the University of Southern California. His research is in the areas of graph signal processing, 3D point cloud processing and compression. His work on point cloud and video compression received best paper awards at IEEE ICIP 2020 and 2022.

Host: Dr. Richard M. Leahy, leahy@usc.edu

Webcast: https://usc.zoom.us/j/95889271695?pwd=TDFRWEsyY1VMWEFFQmZTdkg4ODhYQT09

Location: Hughes Aircraft Electrical Engineering Center (EEB) - 248

WebCast Link: https://usc.zoom.us/j/95889271695?pwd=TDFRWEsyY1VMWEFFQmZTdkg4ODhYQT09

Audiences: Everyone Is Invited

Contact: Mayumi Thrasher