Additional Calendars
Calendar Views
All
Athletics
Conferences and Meetings
Law School
Special Events

ECE Seminar* Speaker: Dr. Marco F. Duarte, UMass Amherst

When: Friday, October 9, 2015
2:00 PM - 4:00 PM
Where: Science & Engineering Building, Lester W. Cory Conference Room: Room 213A
Cost: free
Description: Topic: PARAMETER ESTIMATION IN COMPRESSIVE SENSING

Location: Lester W. Cory Conference Room, Science & Engineering Building (Group II), Room 213A

Abstract:
Compressive sensing (CS) implements simultaneous sensing and compression of sparse and compressible signals based on randomized dimensionality reduction. To recover a signal from its compressive measurements, standard CS algorithms seek the sparsest signal representation in some discrete basis or dictionary that agrees with the measurements. Many applications feature signals that can be represented using a small number of continuous-valued parameters. Such problems have initially been addressed in CS through the design of parametric dictionaries that collect a set of signal observations corresponding to a discretized set of parameter values. These approaches, however, suffer either from resolution limitations due to discretization or from poor performance due to the high coherence of the dictionary, the mismatch between the dictionary and the signal (which may not necessarily be sparse), or both.

This talk will introduce several techniques for compressive parameter estimation (CPE) that aim to alleviate the aforementioned issues, using the time delay estimation and frequency estimation problems common in radar imaging as running examples. First, we use manifold models to characterize the map from parameter space to signal space and employ manifold-based interpolation for parametric dictionaries. In a second approach, we introduce the concept of earth mover's distance for parametric signal representations to directly measure the performance of parameter estimation and leverage it using specially tailored algorithms. Finally, we leverage the connection between approximate message passing methods and denoising algorithms for sparse signals to design compressive parameter estimation approaches based on statistical parameter estimation algorithms. The use of these algorithms allow us to forego the need for a discretized model. We will review the benefits and shortcomings of these proposed alternatives.

Portions of this work are joint with Hamid Dadkhahi, Karsten Fyhn, Dian Mo, and Shermin Hamzehei.

Biography:
Marco F. Duarte is an Assistant Professor in the Department of Electrical and Computer Engineering at the University of Massachusetts Amherst. He received the B.Sc. degree in computer engineering (with distinction) and the M.Sc. degree in electrical engineering from the University of Wisconsin-Madison in 2002 and 2004, respectively, and the Ph.D. degree in electrical engineering from Rice University in 2009. He was an NSF/IPAM Mathematical Sciences Postdoctoral Research Fellow in the Program of Applied and Computational Mathematics at Princeton University from 2009 to 2010, and in the Department of Computer Science at Duke University from 2010 to 2011. His research interests include machine learning, compressed sensing, and sensor networks. Prof. Duarte received the Presidential Fellowship and the Texas Instruments Distinguished Fellowship in 2004 and the Hershel M. Rich Invention Award in 2007, all from Rice University. He coauthored (with C. Hegde and V. Cevher) the Best Student Paper at the 2009 International Workshop on Signal Processing with Adaptive Sparse Structured Representations (SPARS).

The Seminars is open to the public free of charge.

*For further information, please contact Dr. John Buck at 508.999.9237, or via email at jbuck@umassd.edu.
Topical Areas: General Public, University Community, College of Engineering, Electrical and Computer Engineering