A robust approach to independent component analysis of signals with high-level noise measurements

Jianting Cao, Noboru Murata, Shun Ichi Amari, Andrzej Cichocki, Tsunehiro Takeda

    Research output: Contribution to journalArticle

    59 Citations (Scopus)

    Abstract

    In this paper, we propose a robust approach for independent component analysis (ICA) of signals that observations are contaminated with high-level additive noise and/or outliers. The source signals may contain mixtures of both sub-Gaussian and super-Gaussian components, and the number of sources is unknown. Our robust approach includes two procedures. In the first procedure, a robust prewhitening technique is used to reduce the power of additive noise, the dimensionality and the correlation among sources. A cross-validation technique is introduced to estimate the number of sources in this first procedure. In the second procedure, a nonlinear function is derived using the parameterized t-distribution density model. This nonlinear function is robust against the undue influence of outliers fundamentally. Moreover, the stability of the proposed algorithm and the robust property of misestimating the parameters (kurtosis) have been studied. By combining the t-distribution model with a family of light-tailed distributions (sub-Gaussian) model, we can separate the mixture of sub-Gaussian and super-Gaussian source components. Through the analysis of artificially synthesized data and real-world magnetoencephalographic (MEG) data, we illustrate the efficacy of this robust approach.

    Original languageEnglish
    Pages (from-to)631-645
    Number of pages15
    JournalIEEE Transactions on Neural Networks
    Volume14
    Issue number3
    DOIs
    Publication statusPublished - 2003 May

    Fingerprint

    Independent component analysis
    Independent Component Analysis
    Additive noise
    t-distribution
    Additive Noise
    Nonlinear Function
    Outlier
    Kurtosis
    Gaussian Model
    Cross-validation
    Dimensionality
    Efficacy
    Unknown
    Model
    Estimate

    Keywords

    • Cross-validation method
    • Independent component analysis (ICA)
    • Parametric estimation method
    • Principal component analysis (PCA)
    • Robust prewhitening
    • T-distribution density model
    • Unaveraged single-trial MEG data analysis

    ASJC Scopus subject areas

    • Control and Systems Engineering
    • Theoretical Computer Science
    • Electrical and Electronic Engineering
    • Artificial Intelligence
    • Computational Theory and Mathematics
    • Hardware and Architecture

    Cite this

    A robust approach to independent component analysis of signals with high-level noise measurements. / Cao, Jianting; Murata, Noboru; Amari, Shun Ichi; Cichocki, Andrzej; Takeda, Tsunehiro.

    In: IEEE Transactions on Neural Networks, Vol. 14, No. 3, 05.2003, p. 631-645.

    Research output: Contribution to journalArticle

    Cao, Jianting ; Murata, Noboru ; Amari, Shun Ichi ; Cichocki, Andrzej ; Takeda, Tsunehiro. / A robust approach to independent component analysis of signals with high-level noise measurements. In: IEEE Transactions on Neural Networks. 2003 ; Vol. 14, No. 3. pp. 631-645.
    @article{efcefa4fa43849f0a710fcfb6f659c60,
    title = "A robust approach to independent component analysis of signals with high-level noise measurements",
    abstract = "In this paper, we propose a robust approach for independent component analysis (ICA) of signals that observations are contaminated with high-level additive noise and/or outliers. The source signals may contain mixtures of both sub-Gaussian and super-Gaussian components, and the number of sources is unknown. Our robust approach includes two procedures. In the first procedure, a robust prewhitening technique is used to reduce the power of additive noise, the dimensionality and the correlation among sources. A cross-validation technique is introduced to estimate the number of sources in this first procedure. In the second procedure, a nonlinear function is derived using the parameterized t-distribution density model. This nonlinear function is robust against the undue influence of outliers fundamentally. Moreover, the stability of the proposed algorithm and the robust property of misestimating the parameters (kurtosis) have been studied. By combining the t-distribution model with a family of light-tailed distributions (sub-Gaussian) model, we can separate the mixture of sub-Gaussian and super-Gaussian source components. Through the analysis of artificially synthesized data and real-world magnetoencephalographic (MEG) data, we illustrate the efficacy of this robust approach.",
    keywords = "Cross-validation method, Independent component analysis (ICA), Parametric estimation method, Principal component analysis (PCA), Robust prewhitening, T-distribution density model, Unaveraged single-trial MEG data analysis",
    author = "Jianting Cao and Noboru Murata and Amari, {Shun Ichi} and Andrzej Cichocki and Tsunehiro Takeda",
    year = "2003",
    month = "5",
    doi = "10.1109/TNN.2002.806648",
    language = "English",
    volume = "14",
    pages = "631--645",
    journal = "IEEE Transactions on Neural Networks and Learning Systems",
    issn = "2162-237X",
    publisher = "IEEE Computational Intelligence Society",
    number = "3",

    }

    TY - JOUR

    T1 - A robust approach to independent component analysis of signals with high-level noise measurements

    AU - Cao, Jianting

    AU - Murata, Noboru

    AU - Amari, Shun Ichi

    AU - Cichocki, Andrzej

    AU - Takeda, Tsunehiro

    PY - 2003/5

    Y1 - 2003/5

    N2 - In this paper, we propose a robust approach for independent component analysis (ICA) of signals that observations are contaminated with high-level additive noise and/or outliers. The source signals may contain mixtures of both sub-Gaussian and super-Gaussian components, and the number of sources is unknown. Our robust approach includes two procedures. In the first procedure, a robust prewhitening technique is used to reduce the power of additive noise, the dimensionality and the correlation among sources. A cross-validation technique is introduced to estimate the number of sources in this first procedure. In the second procedure, a nonlinear function is derived using the parameterized t-distribution density model. This nonlinear function is robust against the undue influence of outliers fundamentally. Moreover, the stability of the proposed algorithm and the robust property of misestimating the parameters (kurtosis) have been studied. By combining the t-distribution model with a family of light-tailed distributions (sub-Gaussian) model, we can separate the mixture of sub-Gaussian and super-Gaussian source components. Through the analysis of artificially synthesized data and real-world magnetoencephalographic (MEG) data, we illustrate the efficacy of this robust approach.

    AB - In this paper, we propose a robust approach for independent component analysis (ICA) of signals that observations are contaminated with high-level additive noise and/or outliers. The source signals may contain mixtures of both sub-Gaussian and super-Gaussian components, and the number of sources is unknown. Our robust approach includes two procedures. In the first procedure, a robust prewhitening technique is used to reduce the power of additive noise, the dimensionality and the correlation among sources. A cross-validation technique is introduced to estimate the number of sources in this first procedure. In the second procedure, a nonlinear function is derived using the parameterized t-distribution density model. This nonlinear function is robust against the undue influence of outliers fundamentally. Moreover, the stability of the proposed algorithm and the robust property of misestimating the parameters (kurtosis) have been studied. By combining the t-distribution model with a family of light-tailed distributions (sub-Gaussian) model, we can separate the mixture of sub-Gaussian and super-Gaussian source components. Through the analysis of artificially synthesized data and real-world magnetoencephalographic (MEG) data, we illustrate the efficacy of this robust approach.

    KW - Cross-validation method

    KW - Independent component analysis (ICA)

    KW - Parametric estimation method

    KW - Principal component analysis (PCA)

    KW - Robust prewhitening

    KW - T-distribution density model

    KW - Unaveraged single-trial MEG data analysis

    UR - http://www.scopus.com/inward/record.url?scp=0037844871&partnerID=8YFLogxK

    UR - http://www.scopus.com/inward/citedby.url?scp=0037844871&partnerID=8YFLogxK

    U2 - 10.1109/TNN.2002.806648

    DO - 10.1109/TNN.2002.806648

    M3 - Article

    VL - 14

    SP - 631

    EP - 645

    JO - IEEE Transactions on Neural Networks and Learning Systems

    JF - IEEE Transactions on Neural Networks and Learning Systems

    SN - 2162-237X

    IS - 3

    ER -