Title: "Compressive Classification: A Guided Tour"
Speaker: Valerio Cambareri
Location: "Shannon" Seminar Room, Place du Levant 3, Maxwell Building, 1st floor
Date / Time (duration): Thursday 24/09/2015, 14h00 (~45')
Abstract: The mature concept of compressed sensing (CS) is being transferred to the application level as a means to save the physical resources spent in the analog-to-digital interface of challenging signal and image acquisition tasks, i.e., when the underlying sensing process requires a critical amount of time, power, sensor area and cost. For most structured signals, this method amounts to applying a dimensionality-reducing random matrix followed by an accurate, yet computationally expensive recovery algorithm that is capable of producing full-resolution signal recoveries from such low-dimensional measurements.
However, the aforementioned resource saving could be pushed even further if signal acquisition is followed only by simple processing tasks, such as detection or classification of specific events; moreover, this simplification allows one to lighten the burden of computational and accuracy requirements implied by signal recovery.
This talk will provide an overview of two fundamental contributions that tackle the problem of classification from low-dimensional compressive measurements by adopting an intuitive, geometric point of view, moving gradually from classification of a signal within a finite set of known high-dimensional vectors, to the classification of generic signals belonging to disjoint high-dimensional convex bodies such as ellipsoids. The key results will emphasise the role of random projections as a means to provide a universal, non-adaptive dimensionality reduction method while preserving the information required for successful signal classification from compressive measurements.
Valerio Cambareri received the B.S., M.S. (summa cum laude) and Ph.D. degrees in Electronic Engineering from the University of Bologna (Italy) in 2008, 2011 and 2015 respectively. In 2014 he was a Visiting Ph.D. Student in the Integrated Imagers team at IMEC (Belgium). In 2015 he joined the Image and Signal Processing Group of ICTEAM-ELEN, Université catholique de Louvain (Belgium) as a postdoctoral researcher. His current research activity focuses on statistical signal processing, compressed sensing and computational imaging.
Last updated September 29, 2015, at 12:16 PM