Distance metric learning using each category centroid with nuclear norm regularization

Kenta Mikawa, Manabu Kobayashi, Masayuki Goto, Shigeichi Hirasawa

    Research output: Chapter in Book/Report/Conference proceedingConference contribution

    Abstract

    The development in information technology has resulted in more diverse data characteristics and a larger data scale. Therefore, pattern recognition techniques have received significant interest in various fields. In this study, we focus on a pattern recognition technique based on distance metric learning, which is known as the learning method in metric matrix under an arbitrary constraint from the training data. This method can acquire the distance structure, which takes account of the statistical characteristics of the training data. Most distance metric learning methods estimate the metric matrix from pairs of training data. One of the problem of the distance metric learning is that the computational complexity for prediction (i. e. distance calculation) is relatively high especially when the dimension of input data becomes large. To calculate the distance effectively, we propose the way to derive low rank metric matrix with nuclear norm regularization. When solving the optimization problem, we use the alternating direction method of multiplier and proximal gradient. To verify the effectiveness of our proposed method from the viewpoint of classification accuracy and rank reduction, simulation experiments using benchmark data sets are conducted.

    Original languageEnglish
    Title of host publication2017 IEEE Symposium Series on Computational Intelligence, SSCI 2017 - Proceedings
    PublisherInstitute of Electrical and Electronics Engineers Inc.
    Pages1-5
    Number of pages5
    Volume2018-January
    ISBN (Electronic)9781538627259
    DOIs
    Publication statusPublished - 2018 Feb 2
    Event2017 IEEE Symposium Series on Computational Intelligence, SSCI 2017 - Honolulu, United States
    Duration: 2017 Nov 272017 Dec 1

    Other

    Other2017 IEEE Symposium Series on Computational Intelligence, SSCI 2017
    CountryUnited States
    CityHonolulu
    Period17/11/2717/12/1

    Fingerprint

    Distance Metric
    Centroid
    Regularization
    Norm
    Pattern recognition
    Large Data
    Metric
    Pattern Recognition
    Rank Reduction
    Information technology
    Computational complexity
    Method of multipliers
    Alternating Direction Method
    Information Technology
    Simulation Experiment
    Computational Complexity
    Learning
    Benchmark
    Verify
    Gradient

    ASJC Scopus subject areas

    • Artificial Intelligence
    • Computer Science Applications
    • Control and Optimization

    Cite this

    Mikawa, K., Kobayashi, M., Goto, M., & Hirasawa, S. (2018). Distance metric learning using each category centroid with nuclear norm regularization. In 2017 IEEE Symposium Series on Computational Intelligence, SSCI 2017 - Proceedings (Vol. 2018-January, pp. 1-5). Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.1109/SSCI.2017.8280952

    Distance metric learning using each category centroid with nuclear norm regularization. / Mikawa, Kenta; Kobayashi, Manabu; Goto, Masayuki; Hirasawa, Shigeichi.

    2017 IEEE Symposium Series on Computational Intelligence, SSCI 2017 - Proceedings. Vol. 2018-January Institute of Electrical and Electronics Engineers Inc., 2018. p. 1-5.

    Research output: Chapter in Book/Report/Conference proceedingConference contribution

    Mikawa, K, Kobayashi, M, Goto, M & Hirasawa, S 2018, Distance metric learning using each category centroid with nuclear norm regularization. in 2017 IEEE Symposium Series on Computational Intelligence, SSCI 2017 - Proceedings. vol. 2018-January, Institute of Electrical and Electronics Engineers Inc., pp. 1-5, 2017 IEEE Symposium Series on Computational Intelligence, SSCI 2017, Honolulu, United States, 17/11/27. https://doi.org/10.1109/SSCI.2017.8280952
    Mikawa K, Kobayashi M, Goto M, Hirasawa S. Distance metric learning using each category centroid with nuclear norm regularization. In 2017 IEEE Symposium Series on Computational Intelligence, SSCI 2017 - Proceedings. Vol. 2018-January. Institute of Electrical and Electronics Engineers Inc. 2018. p. 1-5 https://doi.org/10.1109/SSCI.2017.8280952
    Mikawa, Kenta ; Kobayashi, Manabu ; Goto, Masayuki ; Hirasawa, Shigeichi. / Distance metric learning using each category centroid with nuclear norm regularization. 2017 IEEE Symposium Series on Computational Intelligence, SSCI 2017 - Proceedings. Vol. 2018-January Institute of Electrical and Electronics Engineers Inc., 2018. pp. 1-5
    @inproceedings{1e81afeb64fb47d8980d1b327ad0cb36,
    title = "Distance metric learning using each category centroid with nuclear norm regularization",
    abstract = "The development in information technology has resulted in more diverse data characteristics and a larger data scale. Therefore, pattern recognition techniques have received significant interest in various fields. In this study, we focus on a pattern recognition technique based on distance metric learning, which is known as the learning method in metric matrix under an arbitrary constraint from the training data. This method can acquire the distance structure, which takes account of the statistical characteristics of the training data. Most distance metric learning methods estimate the metric matrix from pairs of training data. One of the problem of the distance metric learning is that the computational complexity for prediction (i. e. distance calculation) is relatively high especially when the dimension of input data becomes large. To calculate the distance effectively, we propose the way to derive low rank metric matrix with nuclear norm regularization. When solving the optimization problem, we use the alternating direction method of multiplier and proximal gradient. To verify the effectiveness of our proposed method from the viewpoint of classification accuracy and rank reduction, simulation experiments using benchmark data sets are conducted.",
    author = "Kenta Mikawa and Manabu Kobayashi and Masayuki Goto and Shigeichi Hirasawa",
    year = "2018",
    month = "2",
    day = "2",
    doi = "10.1109/SSCI.2017.8280952",
    language = "English",
    volume = "2018-January",
    pages = "1--5",
    booktitle = "2017 IEEE Symposium Series on Computational Intelligence, SSCI 2017 - Proceedings",
    publisher = "Institute of Electrical and Electronics Engineers Inc.",

    }

    TY - GEN

    T1 - Distance metric learning using each category centroid with nuclear norm regularization

    AU - Mikawa, Kenta

    AU - Kobayashi, Manabu

    AU - Goto, Masayuki

    AU - Hirasawa, Shigeichi

    PY - 2018/2/2

    Y1 - 2018/2/2

    N2 - The development in information technology has resulted in more diverse data characteristics and a larger data scale. Therefore, pattern recognition techniques have received significant interest in various fields. In this study, we focus on a pattern recognition technique based on distance metric learning, which is known as the learning method in metric matrix under an arbitrary constraint from the training data. This method can acquire the distance structure, which takes account of the statistical characteristics of the training data. Most distance metric learning methods estimate the metric matrix from pairs of training data. One of the problem of the distance metric learning is that the computational complexity for prediction (i. e. distance calculation) is relatively high especially when the dimension of input data becomes large. To calculate the distance effectively, we propose the way to derive low rank metric matrix with nuclear norm regularization. When solving the optimization problem, we use the alternating direction method of multiplier and proximal gradient. To verify the effectiveness of our proposed method from the viewpoint of classification accuracy and rank reduction, simulation experiments using benchmark data sets are conducted.

    AB - The development in information technology has resulted in more diverse data characteristics and a larger data scale. Therefore, pattern recognition techniques have received significant interest in various fields. In this study, we focus on a pattern recognition technique based on distance metric learning, which is known as the learning method in metric matrix under an arbitrary constraint from the training data. This method can acquire the distance structure, which takes account of the statistical characteristics of the training data. Most distance metric learning methods estimate the metric matrix from pairs of training data. One of the problem of the distance metric learning is that the computational complexity for prediction (i. e. distance calculation) is relatively high especially when the dimension of input data becomes large. To calculate the distance effectively, we propose the way to derive low rank metric matrix with nuclear norm regularization. When solving the optimization problem, we use the alternating direction method of multiplier and proximal gradient. To verify the effectiveness of our proposed method from the viewpoint of classification accuracy and rank reduction, simulation experiments using benchmark data sets are conducted.

    UR - http://www.scopus.com/inward/record.url?scp=85046154292&partnerID=8YFLogxK

    UR - http://www.scopus.com/inward/citedby.url?scp=85046154292&partnerID=8YFLogxK

    U2 - 10.1109/SSCI.2017.8280952

    DO - 10.1109/SSCI.2017.8280952

    M3 - Conference contribution

    AN - SCOPUS:85046154292

    VL - 2018-January

    SP - 1

    EP - 5

    BT - 2017 IEEE Symposium Series on Computational Intelligence, SSCI 2017 - Proceedings

    PB - Institute of Electrical and Electronics Engineers Inc.

    ER -