An Extension to the Natural Gradient Algorithm for Robust Independent Component Analysis in the Presence of Outliers

Muhammad TUFAIL; Masahide ABE; Masayuki KAWAMATA

doi:10.1093/ietfec/e89-a.9.2429

An Extension to the Natural Gradient Algorithm for Robust Independent Component Analysis in the Presence of Outliers

Muhammad TUFAIL, Masahide ABE, Masayuki KAWAMATA

Full Text Views

0

Share
Cite this

Summary :

In this paper, we propose to employ an extension to the natural gradient algorithm for robust Independent Component Analysis against outliers. The standard natural gradient algorithm does not exhibit this property since it employs nonrobust sample estimates for computing higher order moments. In order to overcome this drawback, we propose to use robust alternatives to higher order moments, which are comparatively less sensitive to outliers in the observed data. Some computer simulations are presented to show that the proposed method, as compared to the standard natural gradient algorithm, gives better performance in the presence of outlying data.

Publication: IEICE TRANSACTIONS on Fundamentals Vol.E89-A No.9 pp.2429-2432

Publication Date: 2006/09/01

Publicized

Online ISSN: 1745-1337

DOI: 10.1093/ietfec/e89-a.9.2429

Type of Manuscript: LETTER

Category: Digital Signal Processing

Cite this

Copy

Muhammad TUFAIL, Masahide ABE, Masayuki KAWAMATA, "An Extension to the Natural Gradient Algorithm for Robust Independent Component Analysis in the Presence of Outliers" in IEICE TRANSACTIONS on Fundamentals, vol. E89-A, no. 9, pp. 2429-2432, September 2006, doi: 10.1093/ietfec/e89-a.9.2429.
Abstract: In this paper, we propose to employ an extension to the natural gradient algorithm for robust Independent Component Analysis against outliers. The standard natural gradient algorithm does not exhibit this property since it employs nonrobust sample estimates for computing higher order moments. In order to overcome this drawback, we propose to use robust alternatives to higher order moments, which are comparatively less sensitive to outliers in the observed data. Some computer simulations are presented to show that the proposed method, as compared to the standard natural gradient algorithm, gives better performance in the presence of outlying data.
URL: https://globals.ieice.org/en_transactions/fundamentals/10.1093/ietfec/e89-a.9.2429/_p

Copy

@ARTICLE{e89-a_9_2429,
author={Muhammad TUFAIL, Masahide ABE, Masayuki KAWAMATA, },
journal={IEICE TRANSACTIONS on Fundamentals},
title={An Extension to the Natural Gradient Algorithm for Robust Independent Component Analysis in the Presence of Outliers},
year={2006},
volume={E89-A},
number={9},
pages={2429-2432},
abstract={In this paper, we propose to employ an extension to the natural gradient algorithm for robust Independent Component Analysis against outliers. The standard natural gradient algorithm does not exhibit this property since it employs nonrobust sample estimates for computing higher order moments. In order to overcome this drawback, we propose to use robust alternatives to higher order moments, which are comparatively less sensitive to outliers in the observed data. Some computer simulations are presented to show that the proposed method, as compared to the standard natural gradient algorithm, gives better performance in the presence of outlying data.},
keywords={},
doi={10.1093/ietfec/e89-a.9.2429},
ISSN={1745-1337},
month={September},}

Copy

TY - JOUR
TI - An Extension to the Natural Gradient Algorithm for Robust Independent Component Analysis in the Presence of Outliers
T2 - IEICE TRANSACTIONS on Fundamentals
SP - 2429
EP - 2432
AU - Muhammad TUFAIL
AU - Masahide ABE
AU - Masayuki KAWAMATA
PY - 2006
DO - 10.1093/ietfec/e89-a.9.2429
JO - IEICE TRANSACTIONS on Fundamentals
SN - 1745-1337
VL - E89-A
IS - 9
JA - IEICE TRANSACTIONS on Fundamentals
Y1 - September 2006
AB - In this paper, we propose to employ an extension to the natural gradient algorithm for robust Independent Component Analysis against outliers. The standard natural gradient algorithm does not exhibit this property since it employs nonrobust sample estimates for computing higher order moments. In order to overcome this drawback, we propose to use robust alternatives to higher order moments, which are comparatively less sensitive to outliers in the observed data. Some computer simulations are presented to show that the proposed method, as compared to the standard natural gradient algorithm, gives better performance in the presence of outlying data.
ER -