The study of sparsity in data-rich applications has recently garnered significant attention in the signal processing, statistics, and machine learning communities. Generally speaking, sparsity quantifies the phenomenon where, in a large collection of data, only a small fraction is significant. Techniques that exploit sparsity are currently utilized in a wide variety of applications, including conventional and medical imaging, RF communications, signals intelligence, and bioinformatics, to name a few.

Recently a renewed emphasis has been placed on identifying techniques that make judicious use of sensing resources, with the goal of extracting only the relevant information in high-dimensional but sparse signals. Among these techniques, the most promising approaches typically employ some form of active sampling, or sampling with feedback. In this talk I will discuss one such technique, called Distilled Sensing, and show how active sampling can result in rather surprising (and dramatic) improvements, enabling the detection or estimation of sparse signals that are otherwise unrecoverable using traditional sampling methods.

Host: Richard Baraniuk

Biography of Jarvis Haupt:
Jarvis Haupt received the B.S. in Electrical Engineering, with an additional major in Mathematics, in 2002 from the University of Wisconsin - Madison, where he graduated with highest distinction. Remaining at Wisconsin, he pursued a dual-focus in Communications/Signal Processing and Computer Architecture, and completed the M.S. in Electrical Engineering in 2003. He continued at UW-Madison, where he began working with Professor Robert Nowak on sparse recovery problems. He has made several contributions to the emerging field of Compressed Sensing and was an active participant in the DARPA Analog-to-Information program, and his recent work on active sampling methods provides new insight into the fundamental limits of sparse recovery. He received the Ph.D. degree in Electrical Engineering, with a minor in Mathematics, in August 2009.

Dr. Haupt has completed internships at Georgia Pacific, Domtar Industries, Cray, and L-3 Communications. He is a recipient of several academic awards, including the Wisconsin Academic Excellence Scholarship, the Ford Motor Company Scholarship, the Consolidated Papers Tuition Scholarship, the Frank D. Cady Mathematics Scholarship, and the Claude and Dora Richardson Distinguished Fellowship. He was Chair of the College of Engineering Teaching Improvement Program at the University of Wisconsin - Madison for two semesters, and received Honorable Mention for the Gerald Holdridge Teaching Award for his work as a teaching assistant. His interests broadly include statistical signal processing and learning theory, computer architecture, number theory, and economic policy. He is also a Certified Professional Locksmith.