Compressed Sensing with Redundant Dictionaries

Τοποθεσία: Αίθουσα Α1, Ώρα: 14:00

Τίτλος: Compressed Sensing with Redundant Dictionaries

Ομιλητής: Yannis Kopsinis

Abstract:

Compressed Sensing (CS) provides a systematic framework that allows the estimation of an unknown signal or system based on a number of observations which is significantly reduced compared to the signal/system dimension. It exploits the fact that most of the signals and systems of interest, e.g., medical signals, photos, voice signals, telecommunication channels, etc. are sparse, i.e., they live in low dimensional subspaces. However, conventional CS fails to provide recovery guarantees in the case of high practical interest which concern signals, sparse in dictionaries which are redundant. In this talk, CS for signals which are sparse in orthonormal basis and redundant dictionaries will be presented and the disadvantages of conventional CS will be pointed out. Moreover, the potential of promising CS variants that appeared very recently such as the Analysis Model-based CS, Cosparsity promoting CS and Signal space CS will be discussed. Throughout the presentation, the emphasis is given to the geometrical interpretation of the theorems and methods.

Short Bio:

Yannis Kopsinis (M'04) received the B.Sc, (1998) and the Ph.D. (2003) degrees from the Department of Informatics and Telecommunications, University of Athens. From January 2004 to December 2005, he was a research fellow with the Institute for Digital Communications, School of Engineering and Electronics, the University of Edinburgh, U.K. From January 2006 to September 2009, he was a senior researcher in the same department. He is currently a Ramon Y. Cajal fellow in the University of Granada, Spain. He recently got a Marie Curie grand for conducting research on the topic of online sparsity aware learning with Prof Theodoridis in the Department of Informatics and Telecommunications, UoA. His current research interests include adaptive signal processing, time-frequency analysis, and compressed sensing.