Date/Time: Tuesday, June 30, 2:00pm-2:40pm
Room: CavourSlide set: PDF (46Mb)
The slides of the talk are now available here.
In many applications such as social, sensor, and neuronal networks or multimedia analytics, high-dimensional data naturally reside on the vertices of weighted graphs. The emerging field of signal processing on graphs merges algebraic and spectral graph theoretic concepts with computational harmonic analysis to process such signals on graphs. In this short overview, we outline the main challenges of the area, discuss different ways to define graph spectral domains, which are the analogs to the classical frequency domain, and highlight the importance of incorporating the irregular structures of graph data domains when processing signals on graphs. We then review methods to generalize fundamental operators such as filtering, translation, transforms and dictionaries to the graph setting. We finally show illustrative applications of this emerging framework to challenging multimedia processing problems.
SpeakerPascal Frossard, EPFL
Pascal Frossard has been on the EE faculty of EPFL since 2003, where he heads the Signal Processing Laboratory (LTS4). His research interests include graph signal processing, image representation and coding, visual information analysis, and distributed signal processing and communications. Before joining EPFL, he was a member of the research staff at the IBM T. J. Watson Research Center, Yorktown Heights, NY, where he worked on media coding and streaming technologies. He has served as an Associate or Guest Editor in numerous leading journals in the multimedia community, and has been a member of the organizing committee of major multimedia conferences. He received the Swiss NSF Professorship Award in 2003, the IBM Faculty Award in 2005, the IBM Exploratory Stream Analytics Innovation Award in 2008 and the IEEE Transactions on Multimedia Best Paper Award in 2011.