Expectation–Maximization approaches to independent component analysis

Mingjun Zhong; Huanwen Tang; Yiyuan Tang

doi:10.1016/j.neucom.2004.06.007

Expectation–Maximization approaches to independent component analysis

Mingjun Zhong, Huanwen Tang, Yiyuan Tang

School of Informatics

Research output: Contribution to journal › Article › peer-review

Abstract

Expectation–Maximization (EM) algorithms for independent component analysis are presented in this paper. For super-Gaussian sources, a variational method is employed to develop an EM algorithm in closed form for learning the mixing matrix and inferring the independent components. For sub-Gaussian sources, a symmetrical form of the Pearson mixture model (Neural Comput. 11 (2) (1999) 417–441) is used as the prior, which also enables the development of an EM algorithm in fclosed form for parameter estimation.

Original language	English
Pages (from-to)	503-512
Number of pages	10
Journal	Neurocomputing
Volume	61
DOIs	https://doi.org/10.1016/j.neucom.2004.06.007
Publication status	Published - Oct 2004

Keywords / Materials (for Non-textual outputs)

Independent component analysis
Overcomplete representations
EM algorithm
Variational method

Access to Document

10.1016/j.neucom.2004.06.007

http://www.sciencedirect.com/science/article/pii/S0925231204003121

Cite this

@article{4f515d91dc1643718fab065cb57c5ad4,

title = "Expectation–Maximization approaches to independent component analysis",

abstract = "Expectation–Maximization (EM) algorithms for independent component analysis are presented in this paper. For super-Gaussian sources, a variational method is employed to develop an EM algorithm in closed form for learning the mixing matrix and inferring the independent components. For sub-Gaussian sources, a symmetrical form of the Pearson mixture model (Neural Comput. 11 (2) (1999) 417–441) is used as the prior, which also enables the development of an EM algorithm in fclosed form for parameter estimation.",

keywords = "Independent component analysis, Overcomplete representations, EM algorithm, Variational method",

author = "Mingjun Zhong and Huanwen Tang and Yiyuan Tang",

year = "2004",

month = oct,

doi = "10.1016/j.neucom.2004.06.007",

language = "English",

volume = "61",

pages = "503--512",

journal = "Neurocomputing",

issn = "0925-2312",

publisher = "Elsevier",

}

TY - JOUR

T1 - Expectation–Maximization approaches to independent component analysis

AU - Zhong, Mingjun

AU - Tang, Huanwen

AU - Tang, Yiyuan

PY - 2004/10

Y1 - 2004/10

N2 - Expectation–Maximization (EM) algorithms for independent component analysis are presented in this paper. For super-Gaussian sources, a variational method is employed to develop an EM algorithm in closed form for learning the mixing matrix and inferring the independent components. For sub-Gaussian sources, a symmetrical form of the Pearson mixture model (Neural Comput. 11 (2) (1999) 417–441) is used as the prior, which also enables the development of an EM algorithm in fclosed form for parameter estimation.

AB - Expectation–Maximization (EM) algorithms for independent component analysis are presented in this paper. For super-Gaussian sources, a variational method is employed to develop an EM algorithm in closed form for learning the mixing matrix and inferring the independent components. For sub-Gaussian sources, a symmetrical form of the Pearson mixture model (Neural Comput. 11 (2) (1999) 417–441) is used as the prior, which also enables the development of an EM algorithm in fclosed form for parameter estimation.

KW - Independent component analysis

KW - Overcomplete representations

KW - EM algorithm

KW - Variational method

U2 - 10.1016/j.neucom.2004.06.007

DO - 10.1016/j.neucom.2004.06.007

M3 - Article

SN - 0925-2312

VL - 61

SP - 503

EP - 512

JO - Neurocomputing

JF - Neurocomputing

ER -

Expectation–Maximization approaches to independent component analysis

Abstract

Keywords / Materials (for Non-textual outputs)

Access to Document

Fingerprint

Cite this