Distance metric learning based on different ℓ1 regularized metric matrices in each category

Kenta Mikawa, Manabu Kobayashi, Masayuki Goto, Shigeichi Hirasawa

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

    Abstract

    The distance metric learning is the method to learn the distance metric from training data considering its statistical characteristics under the arbitrary constraints. To obtain the desirable distance metric, the optimization problem is solved. Most of the distance metric learning methods aim to gain the global optimal metric matrix. However there is a possibility that the global metric matrix cannot express the statistical characteristics of each category in detail. In addition, if the dimension of input data increase, the computational cost of calculating distance between data increases either. To avoid this problem, we adopt the way to use the l1 regularization to gain sparse metric matrix. By combining those, we focus on the way to deriving the plural metric matrices with a sparse structure in this study. To verify the effective ness of our proposed method, we conduct simulation experiments by using UCI machine learning repository.

    Original languageEnglish
    Title of host publicationProceedings of 2016 International Symposium on Information Theory and Its Applications, ISITA 2016
    PublisherInstitute of Electrical and Electronics Engineers Inc.
    Pages285-289
    Number of pages5
    ISBN (Electronic)9784885523090
    Publication statusPublished - 2017 Feb 2
    Event3rd International Symposium on Information Theory and Its Applications, ISITA 2016 - Monterey, United States
    Duration: 2016 Oct 302016 Nov 2

    Other

    Other3rd International Symposium on Information Theory and Its Applications, ISITA 2016
    CountryUnited States
    CityMonterey
    Period16/10/3016/11/2

    Fingerprint

    distance learning
    learning method
    Learning systems
    simulation
    experiment
    costs
    learning
    Costs
    Experiments

    ASJC Scopus subject areas

    • Computer Networks and Communications
    • Hardware and Architecture
    • Information Systems
    • Signal Processing
    • Library and Information Sciences

    Cite this

    Mikawa, K., Kobayashi, M., Goto, M., & Hirasawa, S. (2017). Distance metric learning based on different ℓ1 regularized metric matrices in each category. In Proceedings of 2016 International Symposium on Information Theory and Its Applications, ISITA 2016 (pp. 285-289). [7840431] Institute of Electrical and Electronics Engineers Inc..

    Distance metric learning based on different ℓ1 regularized metric matrices in each category. / Mikawa, Kenta; Kobayashi, Manabu; Goto, Masayuki; Hirasawa, Shigeichi.

    Proceedings of 2016 International Symposium on Information Theory and Its Applications, ISITA 2016. Institute of Electrical and Electronics Engineers Inc., 2017. p. 285-289 7840431.

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

    Mikawa, K, Kobayashi, M, Goto, M & Hirasawa, S 2017, Distance metric learning based on different ℓ1 regularized metric matrices in each category. in Proceedings of 2016 International Symposium on Information Theory and Its Applications, ISITA 2016., 7840431, Institute of Electrical and Electronics Engineers Inc., pp. 285-289, 3rd International Symposium on Information Theory and Its Applications, ISITA 2016, Monterey, United States, 16/10/30.
    Mikawa K, Kobayashi M, Goto M, Hirasawa S. Distance metric learning based on different ℓ1 regularized metric matrices in each category. In Proceedings of 2016 International Symposium on Information Theory and Its Applications, ISITA 2016. Institute of Electrical and Electronics Engineers Inc. 2017. p. 285-289. 7840431
    Mikawa, Kenta ; Kobayashi, Manabu ; Goto, Masayuki ; Hirasawa, Shigeichi. / Distance metric learning based on different ℓ1 regularized metric matrices in each category. Proceedings of 2016 International Symposium on Information Theory and Its Applications, ISITA 2016. Institute of Electrical and Electronics Engineers Inc., 2017. pp. 285-289
    @inproceedings{6d0d12e83ac347219b7f660309cac64f,
    title = "Distance metric learning based on different ℓ1 regularized metric matrices in each category",
    abstract = "The distance metric learning is the method to learn the distance metric from training data considering its statistical characteristics under the arbitrary constraints. To obtain the desirable distance metric, the optimization problem is solved. Most of the distance metric learning methods aim to gain the global optimal metric matrix. However there is a possibility that the global metric matrix cannot express the statistical characteristics of each category in detail. In addition, if the dimension of input data increase, the computational cost of calculating distance between data increases either. To avoid this problem, we adopt the way to use the l1 regularization to gain sparse metric matrix. By combining those, we focus on the way to deriving the plural metric matrices with a sparse structure in this study. To verify the effective ness of our proposed method, we conduct simulation experiments by using UCI machine learning repository.",
    author = "Kenta Mikawa and Manabu Kobayashi and Masayuki Goto and Shigeichi Hirasawa",
    year = "2017",
    month = "2",
    day = "2",
    language = "English",
    pages = "285--289",
    booktitle = "Proceedings of 2016 International Symposium on Information Theory and Its Applications, ISITA 2016",
    publisher = "Institute of Electrical and Electronics Engineers Inc.",
    address = "United States",

    }

    TY - GEN

    T1 - Distance metric learning based on different ℓ1 regularized metric matrices in each category

    AU - Mikawa, Kenta

    AU - Kobayashi, Manabu

    AU - Goto, Masayuki

    AU - Hirasawa, Shigeichi

    PY - 2017/2/2

    Y1 - 2017/2/2

    N2 - The distance metric learning is the method to learn the distance metric from training data considering its statistical characteristics under the arbitrary constraints. To obtain the desirable distance metric, the optimization problem is solved. Most of the distance metric learning methods aim to gain the global optimal metric matrix. However there is a possibility that the global metric matrix cannot express the statistical characteristics of each category in detail. In addition, if the dimension of input data increase, the computational cost of calculating distance between data increases either. To avoid this problem, we adopt the way to use the l1 regularization to gain sparse metric matrix. By combining those, we focus on the way to deriving the plural metric matrices with a sparse structure in this study. To verify the effective ness of our proposed method, we conduct simulation experiments by using UCI machine learning repository.

    AB - The distance metric learning is the method to learn the distance metric from training data considering its statistical characteristics under the arbitrary constraints. To obtain the desirable distance metric, the optimization problem is solved. Most of the distance metric learning methods aim to gain the global optimal metric matrix. However there is a possibility that the global metric matrix cannot express the statistical characteristics of each category in detail. In addition, if the dimension of input data increase, the computational cost of calculating distance between data increases either. To avoid this problem, we adopt the way to use the l1 regularization to gain sparse metric matrix. By combining those, we focus on the way to deriving the plural metric matrices with a sparse structure in this study. To verify the effective ness of our proposed method, we conduct simulation experiments by using UCI machine learning repository.

    UR - http://www.scopus.com/inward/record.url?scp=85015158705&partnerID=8YFLogxK

    UR - http://www.scopus.com/inward/citedby.url?scp=85015158705&partnerID=8YFLogxK

    M3 - Conference contribution

    AN - SCOPUS:85015158705

    SP - 285

    EP - 289

    BT - Proceedings of 2016 International Symposium on Information Theory and Its Applications, ISITA 2016

    PB - Institute of Electrical and Electronics Engineers Inc.

    ER -