An adaptive nonlinear function controlled by kurtosis for blind source separation

フォーマット:

論文

責任表示:

Nakayama, Kenji ; Hirano, Akihiro ; Sakai, T.

言語:

英語

出版情報:

Institute of Electrical and Electronics Engineers (IEEE), 2002-05-01

著者名:

• Nakayama, Kenji
• Hirano, Akihiro
• Sakai, T.

掲載情報:

Proceedings of the International Joint Conference on Neural Networks

巻:

26

開始ページ:

1234

終了ページ:

1239

バージョン:

publisher

概要:

金沢大学大学院自然科学研究科情報システム<br />In blind source separation, convergence and separation performances are highly dependent on a relation between probability density functions (pdf) of signal sources and nonlinear functions used in updating co … efficients of a separation block. This relation was analyzed based on kurtosis κ4. It was suggested that tanh y and y3, where y is the output, are useful nonlinear functions for super-Gaussian (κ4 > 0) and sub-Gaussian (κ4 < 0), respectively. In this paper, an adaptive nonlinear function is proposed. It has a form of f(y) = a tanh y + (1 - a)y3/4, where a is controlled by the kurtosis of the output signal yκ(n). It is assumed that the pdf p(y) of the output signal satisfies the stability condition f(y) = -(dp(y)/dy)/p(y). Based on this assumption, the parameter a and the kurtosis is related. This relation approximated by a function a = q(κ4). In a learning process, κ4(n) of the output signal is calculated at each sample n, and a(n) is determined by a(n) = q(κ4(n)). Then, the nonlinear function f(y) is adjusted. Blind separation of music signals of 2-5 channels were simulated. The proposed method is superior to a method, which switches tanh y and y3 based on polarity of κ4(n). 続きを見る

URL:

http://hdl.handle.net/2297/6886