A sequential pattern classifier based on hidden Markov kernel machine and its application to phoneme classification

Yotaro Kubo, Shinji Watanabe, Atsushi Nakamura, Erik McDermott, Tetsunori Kobayashi

    Research output: Contribution to journalArticle

    7 Citations (Scopus)

    Abstract

    This paper describes a novel classifier for sequential data based on nonlinear classification derived from kernel methods. In the proposed method, kernel methods are used for enhancing the emission probability density functions (pdfs) of hidden Markov models (HMMs). Because the emission pdfs enhanced by kernel methods have sufficient nonlinear classification performance, mixture models such as Gaussian mixture models (GMMs), which might cause problems of overfitting and local optima, are not necessary in the proposed method. Unlike the methods used in earlier studies on sequential pattern classification using kernel methods, our method can be regarded as an extension of conventional HMMs, and therefore, it can completely model the transition of hidden states with the observed vectors. Therefore, our method can be applied to many applications developed with conventional HMMs, especially for speech recognition. In this paper, we carried out an isolated phoneme classification as a preliminary experiment in order to evaluate the efficiency of the proposed sequential pattern classifier. We confirmed that the proposed method achieved steady improvements as compared to conventional HMMs with Gaussian-mixture emission pdfs trained by the maximum likelihood and the maximum mutual information procedures.

    Original languageEnglish
    Article number5570878
    Pages (from-to)974-984
    Number of pages11
    JournalIEEE Journal on Selected Topics in Signal Processing
    Volume4
    Issue number6
    DOIs
    Publication statusPublished - 2010 Dec

    Fingerprint

    Hidden Markov models
    Classifiers
    Probability density function
    Speech recognition
    Maximum likelihood
    Pattern recognition
    Experiments

    Keywords

    • Discriminative training
    • hidden Markov models (HMMs)
    • kernel methods
    • sequential pattern classifiers

    ASJC Scopus subject areas

    • Electrical and Electronic Engineering
    • Signal Processing

    Cite this

    A sequential pattern classifier based on hidden Markov kernel machine and its application to phoneme classification. / Kubo, Yotaro; Watanabe, Shinji; Nakamura, Atsushi; McDermott, Erik; Kobayashi, Tetsunori.

    In: IEEE Journal on Selected Topics in Signal Processing, Vol. 4, No. 6, 5570878, 12.2010, p. 974-984.

    Research output: Contribution to journalArticle

    @article{2af7112da8494667a9a4663983e66f48,
    title = "A sequential pattern classifier based on hidden Markov kernel machine and its application to phoneme classification",
    abstract = "This paper describes a novel classifier for sequential data based on nonlinear classification derived from kernel methods. In the proposed method, kernel methods are used for enhancing the emission probability density functions (pdfs) of hidden Markov models (HMMs). Because the emission pdfs enhanced by kernel methods have sufficient nonlinear classification performance, mixture models such as Gaussian mixture models (GMMs), which might cause problems of overfitting and local optima, are not necessary in the proposed method. Unlike the methods used in earlier studies on sequential pattern classification using kernel methods, our method can be regarded as an extension of conventional HMMs, and therefore, it can completely model the transition of hidden states with the observed vectors. Therefore, our method can be applied to many applications developed with conventional HMMs, especially for speech recognition. In this paper, we carried out an isolated phoneme classification as a preliminary experiment in order to evaluate the efficiency of the proposed sequential pattern classifier. We confirmed that the proposed method achieved steady improvements as compared to conventional HMMs with Gaussian-mixture emission pdfs trained by the maximum likelihood and the maximum mutual information procedures.",
    keywords = "Discriminative training, hidden Markov models (HMMs), kernel methods, sequential pattern classifiers",
    author = "Yotaro Kubo and Shinji Watanabe and Atsushi Nakamura and Erik McDermott and Tetsunori Kobayashi",
    year = "2010",
    month = "12",
    doi = "10.1109/JSTSP.2010.2076030",
    language = "English",
    volume = "4",
    pages = "974--984",
    journal = "IEEE Journal on Selected Topics in Signal Processing",
    issn = "1932-4553",
    publisher = "Institute of Electrical and Electronics Engineers Inc.",
    number = "6",

    }

    TY - JOUR

    T1 - A sequential pattern classifier based on hidden Markov kernel machine and its application to phoneme classification

    AU - Kubo, Yotaro

    AU - Watanabe, Shinji

    AU - Nakamura, Atsushi

    AU - McDermott, Erik

    AU - Kobayashi, Tetsunori

    PY - 2010/12

    Y1 - 2010/12

    N2 - This paper describes a novel classifier for sequential data based on nonlinear classification derived from kernel methods. In the proposed method, kernel methods are used for enhancing the emission probability density functions (pdfs) of hidden Markov models (HMMs). Because the emission pdfs enhanced by kernel methods have sufficient nonlinear classification performance, mixture models such as Gaussian mixture models (GMMs), which might cause problems of overfitting and local optima, are not necessary in the proposed method. Unlike the methods used in earlier studies on sequential pattern classification using kernel methods, our method can be regarded as an extension of conventional HMMs, and therefore, it can completely model the transition of hidden states with the observed vectors. Therefore, our method can be applied to many applications developed with conventional HMMs, especially for speech recognition. In this paper, we carried out an isolated phoneme classification as a preliminary experiment in order to evaluate the efficiency of the proposed sequential pattern classifier. We confirmed that the proposed method achieved steady improvements as compared to conventional HMMs with Gaussian-mixture emission pdfs trained by the maximum likelihood and the maximum mutual information procedures.

    AB - This paper describes a novel classifier for sequential data based on nonlinear classification derived from kernel methods. In the proposed method, kernel methods are used for enhancing the emission probability density functions (pdfs) of hidden Markov models (HMMs). Because the emission pdfs enhanced by kernel methods have sufficient nonlinear classification performance, mixture models such as Gaussian mixture models (GMMs), which might cause problems of overfitting and local optima, are not necessary in the proposed method. Unlike the methods used in earlier studies on sequential pattern classification using kernel methods, our method can be regarded as an extension of conventional HMMs, and therefore, it can completely model the transition of hidden states with the observed vectors. Therefore, our method can be applied to many applications developed with conventional HMMs, especially for speech recognition. In this paper, we carried out an isolated phoneme classification as a preliminary experiment in order to evaluate the efficiency of the proposed sequential pattern classifier. We confirmed that the proposed method achieved steady improvements as compared to conventional HMMs with Gaussian-mixture emission pdfs trained by the maximum likelihood and the maximum mutual information procedures.

    KW - Discriminative training

    KW - hidden Markov models (HMMs)

    KW - kernel methods

    KW - sequential pattern classifiers

    UR - http://www.scopus.com/inward/record.url?scp=78649262525&partnerID=8YFLogxK

    UR - http://www.scopus.com/inward/citedby.url?scp=78649262525&partnerID=8YFLogxK

    U2 - 10.1109/JSTSP.2010.2076030

    DO - 10.1109/JSTSP.2010.2076030

    M3 - Article

    VL - 4

    SP - 974

    EP - 984

    JO - IEEE Journal on Selected Topics in Signal Processing

    JF - IEEE Journal on Selected Topics in Signal Processing

    SN - 1932-4553

    IS - 6

    M1 - 5570878

    ER -