Implementation of Infomax ICA Algorithm for Blind Source Separation

Author

Moreno, L. Noe Oliva ; Arce, Miguel A Alemán ; Lamont, Jair García

Author_Institution

Escuela Super. de Comput., Inst. Politec. Nac., Mexico City

fYear

2008

fDate

Sept. 30 2008-Oct. 3 2008

Firstpage

447

Lastpage

451

Abstract

In this paper, is use a field programmable gate array (FPGA) to implement an information-maximization (Infomax) algorithm to blind source separation (BSS). In this work, we show the neural network architecture for several inputs and the performance of its hardware implementation. We achieve simulations results similar as estimated in MATLAB. Some simulations results using VHDL are presented using music and voice recorded. The study of performance is made in a net of 2 times 2.

Keywords

blind source separation; field programmable gate arrays; hardware description languages; independent component analysis; mathematics computing; neural net architecture; Infomax ICA algorithm; MATLAB; VHDL; blind source separation; field programmable gate array; information-maximization algorithm; music recording; neural network architecture; voice recording; Biosensors; Blind source separation; Entropy; Field programmable gate arrays; Independent component analysis; MATLAB; Mutual information; Neural networks; Signal processing algorithms; Source separation; BSS; FPGA; ICA; Neural Networks;

fLanguage

English

Publisher

ieee

Conference_Titel

Electronics, Robotics and Automotive Mechanics Conference, 2008. CERMA '08

Conference_Location

Morelos

Print_ISBN

978-0-7695-3320-9

Type

conf

DOI

10.1109/CERMA.2008.37

Filename

4641113