Penalized PARAFAC analysis of spontaneous EEG recordings

Eduardo Martínez-Montes, José M. Sánchez-Bornot, Pedro A. Valdés-Sosa

    Research output: Contribution to journalArticle

    10 Citations (Scopus)

    Abstract

    The multidimensional nature of neuroscience data has made the use of multi-way statistical analysis suitable in this field. Parallel Factor Analysis (PARAFAC) is a multidimensional generalization of PCA with the advantage of offering unique solutions. However, imposing physiologically acceptable constraints would improve the interpretation of this type of analysis. In this work we propose a new algorithm called Alternating Penalized Least Squares to estimate PARAFAC solutions using different kinds of soft penalization. The algorithm relies on the recent generalization of modified Newton-Raphson techniques to estimate a multiple penalized least squares model. Applied to semi-synthetic and real spontaneous EEG time-varying spectra, we show that a wide range of sparse and smooth solutions can be found separately, as well as with these two properties combined. Smoothness is usually desired in spectra, and different sparse scenarios are observed in the temporal evolution of physiological intermittent phenomena. The degree of constraints can be tuned through the weighting parameters, whose optimal values can be chosen by means of the cross-validation and Corcondia measures.

    LanguageEnglish
    Pages1449-1464
    Number of pages16
    JournalStatistica Sinica
    Volume18
    Issue number4
    Publication statusPublished - 1 Oct 2008

    Fingerprint

    Parallel Factor Analysis
    Penalized Least Squares
    Multiway Analysis
    Alternating Least Squares
    Newton-Raphson
    Neuroscience
    Penalization
    Optimal Parameter
    Smooth Solution
    Cross-validation
    Unique Solution
    Estimate
    Statistical Analysis
    Weighting
    Smoothness
    Time-varying
    Scenarios
    Range of data
    Electroencephalogram
    Generalization

    Keywords

    • Dimensionality reduction
    • EEG
    • PARAFAC
    • Penalized regression

    Cite this

    Martínez-Montes, E., Sánchez-Bornot, J. M., & Valdés-Sosa, P. A. (2008). Penalized PARAFAC analysis of spontaneous EEG recordings. Statistica Sinica, 18(4), 1449-1464.
    Martínez-Montes, Eduardo ; Sánchez-Bornot, José M. ; Valdés-Sosa, Pedro A. / Penalized PARAFAC analysis of spontaneous EEG recordings. In: Statistica Sinica. 2008 ; Vol. 18, No. 4. pp. 1449-1464.
    @article{451595b60150424f96fb1b05f8b1af91,
    title = "Penalized PARAFAC analysis of spontaneous EEG recordings",
    abstract = "The multidimensional nature of neuroscience data has made the use of multi-way statistical analysis suitable in this field. Parallel Factor Analysis (PARAFAC) is a multidimensional generalization of PCA with the advantage of offering unique solutions. However, imposing physiologically acceptable constraints would improve the interpretation of this type of analysis. In this work we propose a new algorithm called Alternating Penalized Least Squares to estimate PARAFAC solutions using different kinds of soft penalization. The algorithm relies on the recent generalization of modified Newton-Raphson techniques to estimate a multiple penalized least squares model. Applied to semi-synthetic and real spontaneous EEG time-varying spectra, we show that a wide range of sparse and smooth solutions can be found separately, as well as with these two properties combined. Smoothness is usually desired in spectra, and different sparse scenarios are observed in the temporal evolution of physiological intermittent phenomena. The degree of constraints can be tuned through the weighting parameters, whose optimal values can be chosen by means of the cross-validation and Corcondia measures.",
    keywords = "Dimensionality reduction, EEG, PARAFAC, Penalized regression",
    author = "Eduardo Mart{\'i}nez-Montes and S{\'a}nchez-Bornot, {Jos{\'e} M.} and Vald{\'e}s-Sosa, {Pedro A.}",
    year = "2008",
    month = "10",
    day = "1",
    language = "English",
    volume = "18",
    pages = "1449--1464",
    journal = "Statistica Sinica",
    issn = "1017-0405",
    publisher = "Institute of Statistical Science",
    number = "4",

    }

    Martínez-Montes, E, Sánchez-Bornot, JM & Valdés-Sosa, PA 2008, 'Penalized PARAFAC analysis of spontaneous EEG recordings', Statistica Sinica, vol. 18, no. 4, pp. 1449-1464.

    Penalized PARAFAC analysis of spontaneous EEG recordings. / Martínez-Montes, Eduardo; Sánchez-Bornot, José M.; Valdés-Sosa, Pedro A.

    In: Statistica Sinica, Vol. 18, No. 4, 01.10.2008, p. 1449-1464.

    Research output: Contribution to journalArticle

    TY - JOUR

    T1 - Penalized PARAFAC analysis of spontaneous EEG recordings

    AU - Martínez-Montes, Eduardo

    AU - Sánchez-Bornot, José M.

    AU - Valdés-Sosa, Pedro A.

    PY - 2008/10/1

    Y1 - 2008/10/1

    N2 - The multidimensional nature of neuroscience data has made the use of multi-way statistical analysis suitable in this field. Parallel Factor Analysis (PARAFAC) is a multidimensional generalization of PCA with the advantage of offering unique solutions. However, imposing physiologically acceptable constraints would improve the interpretation of this type of analysis. In this work we propose a new algorithm called Alternating Penalized Least Squares to estimate PARAFAC solutions using different kinds of soft penalization. The algorithm relies on the recent generalization of modified Newton-Raphson techniques to estimate a multiple penalized least squares model. Applied to semi-synthetic and real spontaneous EEG time-varying spectra, we show that a wide range of sparse and smooth solutions can be found separately, as well as with these two properties combined. Smoothness is usually desired in spectra, and different sparse scenarios are observed in the temporal evolution of physiological intermittent phenomena. The degree of constraints can be tuned through the weighting parameters, whose optimal values can be chosen by means of the cross-validation and Corcondia measures.

    AB - The multidimensional nature of neuroscience data has made the use of multi-way statistical analysis suitable in this field. Parallel Factor Analysis (PARAFAC) is a multidimensional generalization of PCA with the advantage of offering unique solutions. However, imposing physiologically acceptable constraints would improve the interpretation of this type of analysis. In this work we propose a new algorithm called Alternating Penalized Least Squares to estimate PARAFAC solutions using different kinds of soft penalization. The algorithm relies on the recent generalization of modified Newton-Raphson techniques to estimate a multiple penalized least squares model. Applied to semi-synthetic and real spontaneous EEG time-varying spectra, we show that a wide range of sparse and smooth solutions can be found separately, as well as with these two properties combined. Smoothness is usually desired in spectra, and different sparse scenarios are observed in the temporal evolution of physiological intermittent phenomena. The degree of constraints can be tuned through the weighting parameters, whose optimal values can be chosen by means of the cross-validation and Corcondia measures.

    KW - Dimensionality reduction

    KW - EEG

    KW - PARAFAC

    KW - Penalized regression

    UR - http://www.scopus.com/inward/record.url?scp=60149110922&partnerID=8YFLogxK

    M3 - Article

    VL - 18

    SP - 1449

    EP - 1464

    JO - Statistica Sinica

    T2 - Statistica Sinica

    JF - Statistica Sinica

    SN - 1017-0405

    IS - 4

    ER -

    Martínez-Montes E, Sánchez-Bornot JM, Valdés-Sosa PA. Penalized PARAFAC analysis of spontaneous EEG recordings. Statistica Sinica. 2008 Oct 1;18(4):1449-1464.