Proceedings of the Korean Information Science Society Conference (한국정보과학회:학술대회논문집)
- 2004.04b
- /
- Pages.721-723
- /
- 2004
- /
- 1598-5164(pISSN)
A Trust-Region ICA algorithm
Trust-Region ICA 알고리듬
- Park, Heeyoul (Department of Computer Science, POSTECH) ;
- Kim, Sookjeong (Department of Computer Science, POSTEC) ;
- Park, Seungjin (Department of Computer Science, POSTECH)
- Published : 2004.04.01
Abstract
A trust-region method is a quite attractive optimization technique. It is, in general, faster than the steepest descent method and is free of a learning rate unlike the gradient-based methods. In addition to its convergence property (between linear and quadratic convergence), ifs stability is always guaranteed, in contrast to the Newton's method. In this paper, we present an efficient implementation of the maximum likelihood independent component analysis (ICA) using the trust-region method, which leads to trust-region-based ICA (TR-ICA) algorithms. The useful behavior of our TR-ICA algorithms is confimed through numerical experimental results.
Keywords