BROWSE
Detail
Conference Paper
Blind Source Separation using Relative Optimization
Cited - time in 
Share 
Share
- Authors
- In Seon Jang, Seung Kwon Beack, Kyeon Gok Kang, Ki Joeong Nam, Seung Jin Choi
- Issue Date
- 2006-06
- Citation
- Westem Pacific Acoustics Conference (WESPAC) 2006, pp.1-6
- Language
- English
- Type
- Conference Paper
- Abstract
- In this paper we present a blind source separation algorithm that jointly exploits the nonstationarity and second order statistics of sources. To estimate a demixing filter, we applied relative gradient method for joint diagonalization. Relative gradient consists of two steps; the relative optimization technique and the gradient method. This method turned out to be faster and to provide better separation performance than conventional gradient types. Experiments for real recording data confirm the high performance of our proposed method.
- KSP Keywords
- High performance, Joint Diagonalization, Optimization techniques, Relative gradient, Second order statistics(SOS), Separation algorithm, blind source separation(BSS), gradient method, separation performance