Presentation 2008-05-23
Proposal of a nonlinear echo canceller that uses Gaussian process
Junichiro TOMITA, Yuzo HIRAI,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) Gaussian process (GP) is a process where the joint distribution over variables is specified completely by the second-order statistics, namely the mean and the covariance. Because the output data is estimated by a training data set and a test data set, it is though that the influence on the convergence characteristic is small. In this paper, we confirm that both of linear echo-path and nonlinear echo-path follow Gaussian distributions. Furthermore, we propose a nonlinear echo-canceller that uses Gaussian process, and analyze the convergence characteristic of ERLE for stationary colored signal to ensure that Gaussian process can cancel the echo.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) Gaussian Process / Nonlinearity / Convergence / Mean function / Covariance function
Paper # NC2008-7
Date of Issue

Conference Information
Committee NC
Conference Date 2008/5/16(1days)
Place (in Japanese) (See Japanese page)
Place (in English)
Topics (in Japanese) (See Japanese page)
Topics (in English)
Chair
Vice Chair
Secretary
Assistant

Paper Information
Registration To Neurocomputing (NC)
Language JPN
Title (in Japanese) (See Japanese page)
Sub Title (in Japanese) (See Japanese page)
Title (in English) Proposal of a nonlinear echo canceller that uses Gaussian process
Sub Title (in English)
Keyword(1) Gaussian Process
Keyword(2) Nonlinearity
Keyword(3) Convergence
Keyword(4) Mean function
Keyword(5) Covariance function
1st Author's Name Junichiro TOMITA
1st Author's Affiliation University of Tsukuba, Graduate School of Systems and Infomation Engineering()
2nd Author's Name Yuzo HIRAI
2nd Author's Affiliation University of Tsukuba, Graduate School of Systems and Infomation Engineering
Date 2008-05-23
Paper # NC2008-7
Volume (vol) vol.108
Number (no) 54
Page pp.pp.-
#Pages 6
Date of Issue