Bounded Component Analysis: A New Source Separation Approach
  • FENS
  • Bounded Component Analysis: A New Source Separation Approach

You are here

Speaker:  Dr. Alper Erdoğan, Koç University, İstanbul

Title:  "Bounded Component Analysis: A New Source Separation Approach"

Date/Time:  Nov 30 Wednesday at 14.40

Location:  FENS L058
Abstract:
Blind Source Separation (BSS)  is a core problem for several machine learning and signal processing applications. We can list MIMO commnuications, functional brain imaging, audio source separation as some representative applications. Over the last two decades, several different algorithms have been proposed for the solution of the BSS problem. Among these, Independent Component Analysis, which exploits the assumption about the mutual statistical independence of sources, stands out as the most successful and most popular approach.

Recently, Bounded Component Analysis (BCA) scheme has been put forward as an alternative BSS scheme which exploits the boundedness of sources to replace the source independence assumption with a weaker assumption. Therefore, under the generic assumption on source boundedness, BCA provides a more general framework for the separation of both dependent and independent sources.

In this talk,  we introduce the Bounded Component Analysis after a brief primer on BSS problem and ICA approach. The introduction will cover the historical perspective as well as the founding assumptions of the BCA approach. The focus will be on the presentation of our  results regarding the construction, analysis and applications of BCA algorithms.
Bio:
Alper T. Erdogan received his B.S. degree from the Middle East Technical University, Ankara, Turkey in 1993, and the M.S. and Ph.D. degrees from Stanford University, Stanford, CA, in 1995 and 1999, respectively. He was a principal Research Engineer in Globespan-Virata Corporation (formerly Excess Bandwidth and Virata Corporations) from September 1999 to November 2001. He joined Electrical and Electronics Engineering Department of Koc University, Istanbul, Turkey in January 2002, where he is currently an Associate Professor.

Dr. Erdogan is the recipient of TUBITAK Career Award (2005), Koc University Werner Von Siemens Excellence Excellence Award (2007) and TUBA GEBIP Outstanding Young Scientist Award (2008) and TUBITAK Young Scientist Encouragement Award(2010) . Dr. Erdogan served as Associate Editor of IEEE Transactions on Signal Processing and he is currently a member of IEEE Signal Processing Theory and Methods Technical Committee. His research interests include wireless, fiber and wireline communications, adaptive signal processing, optimization, system theory and control, and information theory.