Seminar by Prof. Aleksandar Dogandzic

Prof. Aleksandar Dogand�ić, from Iowa State University, will give the lecture titled: "ECME Thresholding Methods for Sparse Signal Reconstruction". It will take place on June 15th, 2010, at Sala de Graus, from 11:30h to 12:30h.


Abstract:  I will describe our probabilistic framework for developing  and interpreting hard thresholding methods for sparse signal reconstruction and present several new methods that are based on this framework.  The measurements follow an underdetermined linear model, where the regression-coefficient vector is a sum of an unknown deterministic sparse signal component and a zero-mean white Gaussian component with an unknown variance. I will first describe an expectation conditional maximization either (ECME) iteration that guarantees convergence to a local maximum of the likelihood function
of the unknown parameters for a given signal sparsity level.  To analyze the reconstruction accuracy, I will introduce the minimum sparse subspace quotient (SSQ), a more flexible measure of the sampling operator than the well-established restricted isometry property (RIP).  We prove that, if the minimum SSQ is sufficiently large, ECME achieves perfect or near-optimal recovery of sparse or approximately sparse signals, respectively.  I will then describe our double overrelaxation (DORE) thresholding scheme for accelerating the ECME iteration. If the signal sparsity level is unknown, we introduce an unconstrained sparsity selection (USS) criterion for its selection and show that, under certain conditions, applying this criterion is equivalent to finding the sparsest solution of the underlying underdetermined linear system. I will describe our automatic double overrelaxation (ADORE) thresholding method that employs the USS criterion to select the signal sparsity level automatically.

Finally, I will present numerical examples where we employ the proposed methods to reconstruct sparse and approximately sparse signals from tomographic projections and compressive samples.

Bio: Aleksandar Dogandzic received the Dipl. Ing. degree (summa cum laude) in Electrical Engineering from the University of Belgrade, Yugoslavia, in 1995, and the M.S. and Ph.D. degrees in electrical engineering and computer science from the University of Illinois at Chicago (UIC) in 1997 and 2001, respectively.

In August 2001, he joined the Department of Electrical and Computer Engineering, Iowa State University (ISU), Ames, where he is currently an Associate Professor.  His research interests are in statistical signal processing: theory and applications.

Dr. Dogandzic received the 2003 Young Author Best Paper Award and 2004 Signal Processing Magazine Best Paper Award, both by the IEEE Signal Processing Society.  In 2006, he received the CAREER Award by the National Science Foundation. At ISU, he was awarded the 2006-2007 Litton Industries Assistant Professorship in Electrical and Computer Engineering.  Dr. Dogandzic is currently an Associate Editor for the IEEE Transactions on Signal Processing and IEEE Signal Processing Letters. He will serve as a general co-chair of the Fourth International Workshop on Computational Advances in Multi-Sensor Adaptive Processing (CAMSAP 2011).