Teaching
Statistical Signal Processing
Michael A. Lexa
5747 Hummingbird St.
Houston TX, 77096-5903 USA
+1 713.721.6374 (voice)
amlexa@gmail.com
Michael Lexa received his B.S. degree from the University of Notre Dame in 1994, his M.S. degree from the University of Colorado at Boulder in 1996, and his Ph.D. degree from Rice University in 2008 all in electrical engineering. From 1996-1998 he served in the U.S. Air Force, Space and Missile Systems Center, Test and Evaluation Directorate, Kirkland AFB NM where he supported test flights for the National Ballistic Missile program. From 1998-1999, he was assigned to the U.S. Air Force, National Air Intelligence Center, Wright-Patterson AFB OH. In 2003, he interned at the MIT Lincoln Laboratory in Lexington, Massachusetts to work on the some of the theoretical aspects of data fusion.
Research Interests / Projects
Research interests and activities lie in the areas of statistical signal processing, specifically detection, estimation, distributed systems, and statistical learning theory.
Distributed Quantization for Classification: The Impact of Structure and Nonparametric Estimation
Distributed signal processing refers to any signal processing task (estimation, detection, source coding, sampling, etc.) under the defining constraint that either the data, the processing units, or both are physically separated. Such systems often have the ability to tackle complex problems using a divide and conquer strategy, but at the same time their structure, that is the distributed nature of the system (and data), severely constrains how data are processed and thus may degrade system performance. To mitigate these structural constraints, partial information can be shared (communicated) among a distributed system's processing units while largely retaining its ability to tackle computationally tough problems. (Think of a wireless sensor network for example.) Knowing what, when, and to whom communication should occur in any given problem are fundamental questions surrounding distributed signal processing.
This project begins to answer these questions by examining simple, distributed quantization systems optimized such that the performance of a downstream classifier is maximized. Even for these conceptually simple communicative systems, suboptimal estimation techniques need to be used to avoid overwhelming computational complexity. Thus, priority is placed on understanding how structure and these estimation strategies affect and balance performance and computational complexity.
Publications
M.A. Lexa and D.H. Johnson, Distributed Structures, Sequential Optimization, and Quantization for Detection. IEEE Trans. Signal Processing, vol. 56, no. 4, pp.1740-1745, Apr 2008.
M.A. Lexa and D.H. Johnson, Joint Optimization of Distributed Broadcast Quantization Systems for Classification. IEEE Data Compression Conference (DCC'07) Mar 2007.
M.A. Lexa, C.J. Rozell, S. Sinanovic and D.H. Johnson, To cooperate or not to cooperate: Detection Strategies in Sensor Networks. IEEE International Conference on Acoustics, Speech, and Signal Processing (ICASSP'04), Apr 2004.
M.A. Lexa and D.H. Johnson, An Information Processing Approach to Distributed Detection. IEEE Workshop on Statistical Signal Processing (SSP'03), Sep 2003.
M.A. Lexa and D.H. Johnson, Optimizing Binary Decision Systems by Manipulating Transmission Intervals. IEEE International Symposium on Signal Processing and Its Applications (ISSPA'03), July 2003.
M.A. Lexa and D.H. Johnson, A New Look at the Informational Gain of Soft Decisions. IEEE International Conference on Acoustics, Speech, and Signal Processing (ICASSP'03), Apr 2003.
M.A. Lexa and D.H. Johnson, Information Processing Ability of Binary Detectors and Block Decoders. IEEE Digital Signal Processing Workshop (DSP'02), 2002.
This material is presented to ensure timely dissemination of scholarly and technical work. Copyright and all rights therein are retained by authors or by other copyright holders. All persons copying this information are expected to adhere to the terms and constraints invoked by each author's copyright. In most cases, these works may not be reposted without the explicit permission of the copyright holder.
Unpublished Manuscripts
M.A. Lexa, Ph.D. Thesis, Sequential Quantization for Classification: The Impact of Structure and Nonparametric Estimates, Aug 2008. Thesis style (pdf) / Report form (pdf) / Errata (pdf)
M.A. Lexa, Useful Facts about the Kullback-Leibler Discrimination Distance, Dec 2004.
M.A. Lexa, Remembering John Napier and His Logarithms, August 2000.