ORAL COMPREHENSIVE EXAM FOR DOCTORAL CANDIDACY BY: Keith D. Gilbert
When: Thursday,
March 3, 2016
9:00 AM
-
11:00 AM
Where: Science & Engineering Building, Lester W. Cory Conference Room: Room 213A
Cost: free
Description: TOPIC: A MULTIPLE ALGORITHM APPROACH TO BLIND SOURCE SEPARATION
LOCATION: Lester W. Cory Conference Room, SENG-213A
ABSTRACT:
The general source separation problem involves estimating a set of sources from a set of observations in which the sources have been combined. The problem takes on several variants depending on what is known about the sources or the mixing system (blindness), the type of mixing system that produces the observations (linearity and delay dependence), and the number of sources relative to the number of unique mixtures (determinedness). The work presented here investigates the use of multiple source separation algorithms running in parallel to produce a composite set of source estimates, and the research focuses on blind separation of mutually, statistically independent sources from a set of linearly independent mixtures. A blind quality of separation (QoS) measure for a single source estimate that does not rely on other source estimates is derived that is based on information contained within the mixture set about the source estimate. This QoS measure is then applied to a competitive, multiple algorithm approach to the source separation problem, both in the instantaneous and convolutive cases, and results for selection-based and blending-based composite source estimates are given.
NOTE: All ECE Graduate Students are ENCOURAGED to attend.
All interested parties are invited to attend. Open to the public.
Advisor: Dr. Karen Payton
Committee Members: Dr. Antonio H. Costa and Dr. John Buck, Department of Electrical & Computer Engineering; Dr. Joel MacAuslan, President, Speech Technology and Applied Research Corporation; and Dr. Richard Goldhor, Owner, Grapevine Software
*For further information, please contact Dr. Karen L. Payton at 508.999.8434, or via email at kpayton@umassd.edu.
LOCATION: Lester W. Cory Conference Room, SENG-213A
ABSTRACT:
The general source separation problem involves estimating a set of sources from a set of observations in which the sources have been combined. The problem takes on several variants depending on what is known about the sources or the mixing system (blindness), the type of mixing system that produces the observations (linearity and delay dependence), and the number of sources relative to the number of unique mixtures (determinedness). The work presented here investigates the use of multiple source separation algorithms running in parallel to produce a composite set of source estimates, and the research focuses on blind separation of mutually, statistically independent sources from a set of linearly independent mixtures. A blind quality of separation (QoS) measure for a single source estimate that does not rely on other source estimates is derived that is based on information contained within the mixture set about the source estimate. This QoS measure is then applied to a competitive, multiple algorithm approach to the source separation problem, both in the instantaneous and convolutive cases, and results for selection-based and blending-based composite source estimates are given.
NOTE: All ECE Graduate Students are ENCOURAGED to attend.
All interested parties are invited to attend. Open to the public.
Advisor: Dr. Karen Payton
Committee Members: Dr. Antonio H. Costa and Dr. John Buck, Department of Electrical & Computer Engineering; Dr. Joel MacAuslan, President, Speech Technology and Applied Research Corporation; and Dr. Richard Goldhor, Owner, Grapevine Software
*For further information, please contact Dr. Karen L. Payton at 508.999.8434, or via email at kpayton@umassd.edu.
Contact:
ECE: Electrical & Computer Engineering Department 508.999.9164 http://www.umassd.edu/engineering/ece/
Topical Areas: General Public, University Community, College of Engineering, Electrical and Computer Engineering