University of Minnesota
Department of Electrical and Computer Engineering
Abstract - In recent years the signal processing, statistics, and machine learning communities have experienced a flurry of research activity aimed at the development of new "non-traditional" sampling, sensing, and inference methods. For example, recent breakthrough results in compressive sampling have shed new light on our understanding of sampling and reconstruction, leading to revolutionary new technologies in application domains ranging from RF communications and surveillance, to conventional and medical imaging, to genomics, and even industrial quality testing. The enabling feature of this new wave of research is the notion that, in many applications of practical interest, the data being collected exhibit some degree of sparsity. Generally speaking, sparsity describes the phenomenon where the salient information in a collection of data can be parsimoniously represented, for example, by only a small subset of the data itself or by a small number of summary coefficients.
In this talk, I will discuss exciting new work which fuses the notion of sparsity with that of adaptivity in sampling. Adaptive sampling refers to data collection procedures designed to actively seek out highly informative data, in contrast with more traditional methods that rely on fixed designs. I will describe how the inclusion of this (often simple) processing step "in the loop" can lead to dramatic improvements over traditional sampling methods where the data collection process is fixed by design, and I will present some new results to quantify and demonstrate the gains that are achievable when adaptively sensing sparse signals in noisy environments.
Biography - Jarvis Haupt received the B.S. (with highest distinction), M.S., and Ph.D. degrees in Electrical Engineering from the University of Wisconsin - Madison in 2002, 2003, and 2009, respectively. From August 2009 - August 2010, he was a Postdoctoral Research Associate in the Department of Electrical and Computer Engineering at Rice University. He joined the Department of Electrical and Computer Engineering at the University of Minnesota as an Assistant Professor in August 2010. His research interests generally include high dimensional statistical inference, sparse recovery, adaptive sampling techniques, statistical signal processing and learning theory, and applications in communications, network science, remote sensing, imaging, and systems biology.