DocumentCode
2766231
Title
On the INFOMAX algorithm for blind signal separation
Author
Xi, Jiangtao ; Chicharo, J.F. ; Tsoi, Ah Chung ; Siu, Wan-chi
Author_Institution
Wollongong Univ., NSW, Australia
Volume
1
fYear
2000
fDate
2000
Firstpage
425
Abstract
This paper provides an analytical examination of the INFOMAX algorithm and establishes its effectiveness for blind signal separation using extensive simulation results. Results obtained show that the INFOMAX is not able to separate signal sources unless signal pre-processing is carried-out whereby the data to train the separating matrix is decorrelated. Further, results also show that if one uses the decorrelation pre-process alone it is able to effectively separate signal sources in many instances. Hence we conclude that the INFOMAX algorithm may not be a useful approach for signal separation
Keywords
array signal processing; decorrelation; gradient methods; maximum entropy methods; statistical analysis; INFOMAX algorithm; blind signal separation; decorrelation; information-maximization approach; instantaneous mixing; sensor array; signal pre-processing; signal source separation; simulation results; Blind source separation; Decorrelation; Entropy; Jacobian matrices; MIMO; Neural networks; Sensor arrays; Signal analysis; Signal mapping; Source separation;
fLanguage
English
Publisher
ieee
Conference_Titel
Signal Processing Proceedings, 2000. WCCC-ICSP 2000. 5th International Conference on
Conference_Location
Beijing
Print_ISBN
0-7803-5747-7
Type
conf
DOI
10.1109/ICOSP.2000.894523
Filename
894523
Link To Document