Patchworking multiple pairwise distances for learning with distance matrices

Ken Takano, Hideitsu Hino*, Yuki Yoshikawa, Noboru Murata

*この研究の対応する著者

研究成果: Conference contribution

1 被引用数 (Scopus)

抄録

A classification framework using only a set of distance matrices is proposed. The proposed algorithm can learn a classifier only from a set of distance matrices or similarity matrices, hence applicable to structured data, which do not have natural vector representation such as time series and graphs. Random forest is used to explore ideal feature representation based on the distance between points defined by a set of given distance matrices. The effectiveness of the proposed method is evaluated through experiments with point process data and graph structured data.

本文言語English
ホスト出版物のタイトルLatent Variable Analysis and Signal Separation - 12th International Conference, LVA/ICA 2015, Proceedings
編集者Zbynĕk Koldovský, Emmanuel Vincent, Arie Yeredor, Petr Tichavský
出版社Springer Verlag
ページ287-294
ページ数8
ISBN(印刷版)9783319224817
DOI
出版ステータスPublished - 2015
イベント12th International Conference on Latent Variable Analysis and Signal Separation, LVA/ICA 2015 - Liberec, Czech Republic
継続期間: 2015 8月 252015 8月 28

出版物シリーズ

名前Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
9237
ISSN(印刷版)0302-9743
ISSN(電子版)1611-3349

Other

Other12th International Conference on Latent Variable Analysis and Signal Separation, LVA/ICA 2015
国/地域Czech Republic
CityLiberec
Period15/8/2515/8/28

ASJC Scopus subject areas

  • 理論的コンピュータサイエンス
  • コンピュータ サイエンス(全般)

フィンガープリント

「Patchworking multiple pairwise distances for learning with distance matrices」の研究トピックを掘り下げます。これらがまとまってユニークなフィンガープリントを構成します。

引用スタイル