Wednesday Afternoon, October 25
MS40
Theory and Applications of Two-Sided Orthogonal Decompositions
Rank-revealing two-sided orthogonal decompositions are important linear
algebra "tools" in applications, such as signal processing, spectroscopy, and
information retrieval, that depend on fast algorithms, updatable decompositions,
and reliable detection of numerical rank. These decompositions emulate "partial"
singular value decompositions (PSVD), and are faster and less computationally
demanding than the full SVD. The speakers will discuss recent advances and
applications of these decompositions in signal processing, low-rank sparse or structured least
squares problems, and information retrieval.
Organizer: Ricardo D. Fierro
California State University, San Marcos
- 1:30 Modifying the ULV Decomposition for An Efficient Array Processing
- Jesse L. Barlow, Leon H. Sibul, and Peter A. Yoon, The Pennsylvania State University
- 2:00 Partial Generalized SVD and its Application in Estimator-Correlator Array Signal Processing
- Hongyuan Zha, The Pennsylvania State University
- 2:30 Fast Two-Sided Orthogonal Decompositions for Sparse or Structured Low-Rank Problems
- Ricardo D. Fierro, Organizer
- 3:00 Low-Rank Orthogonal Decomposition for Information Retrieval Applications
- Michael W. Berry, University of Tennessee and Ricardo D. Fierro, Organizer
7/27/95