Information geometric perspective of modal linear regression

Keishi Sando, Shotaro Akaho, Noboru Murata, Hideitsu Hino

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

    1 Citation (Scopus)

    Abstract

    Modal linear regression (MLR) is a standard method for modeling the conditional mode of a response variable using a linear combination of explanatory variables. It is effective when dealing with response variables with an asymmetric, multi-modal distribution. Because of the nonparametric nature of MLR, it is difficult to construct a statistical model manifold in the sense of information geometry. In this work, a model manifold is constructed using observations instead of explicit parametric models. We also propose a method for constructing a data manifold based on an empirical distribution. The em algorithm, which is a geometric formulation of the EM algorithm, of MLR is shown to be equivalent to the conventional EM algorithm of MLR.

    Original languageEnglish
    Title of host publicationNeural Information Processing - 25th International Conference, ICONIP 2018, Proceedings
    EditorsLong Cheng, Seiichi Ozawa, Andrew Chi Sing Leung
    PublisherSpringer-Verlag
    Pages535-545
    Number of pages11
    ISBN (Print)9783030041816
    DOIs
    Publication statusPublished - 2018 Jan 1
    Event25th International Conference on Neural Information Processing, ICONIP 2018 - Siem Reap, Cambodia
    Duration: 2018 Dec 132018 Dec 16

    Publication series

    NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
    Volume11303 LNCS
    ISSN (Print)0302-9743
    ISSN (Electronic)1611-3349

    Other

    Other25th International Conference on Neural Information Processing, ICONIP 2018
    CountryCambodia
    CitySiem Reap
    Period18/12/1318/12/16

    Keywords

    • EM algorithm
    • Information geometry
    • Modal linear regression

    ASJC Scopus subject areas

    • Theoretical Computer Science
    • Computer Science(all)

    Fingerprint Dive into the research topics of 'Information geometric perspective of modal linear regression'. Together they form a unique fingerprint.

  • Cite this

    Sando, K., Akaho, S., Murata, N., & Hino, H. (2018). Information geometric perspective of modal linear regression. In L. Cheng, S. Ozawa, & A. C. S. Leung (Eds.), Neural Information Processing - 25th International Conference, ICONIP 2018, Proceedings (pp. 535-545). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 11303 LNCS). Springer-Verlag. https://doi.org/10.1007/978-3-030-04182-3_47