Patchworking multiple pairwise distances for learning with distance matrices

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

*Corresponding author for this work

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

1 Citation (Scopus)

Abstract

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.

Original languageEnglish
Title of host publicationLatent Variable Analysis and Signal Separation - 12th International Conference, LVA/ICA 2015, Proceedings
EditorsZbynĕk Koldovský, Emmanuel Vincent, Arie Yeredor, Petr Tichavský
PublisherSpringer Verlag
Pages287-294
Number of pages8
ISBN (Print)9783319224817
DOIs
Publication statusPublished - 2015
Event12th International Conference on Latent Variable Analysis and Signal Separation, LVA/ICA 2015 - Liberec, Czech Republic
Duration: 2015 Aug 252015 Aug 28

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume9237
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Other

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

Keywords

  • Classification
  • Decision trees
  • Graph kernel
  • Random forest
  • Spike train
  • Structured data

ASJC Scopus subject areas

  • Theoretical Computer Science
  • Computer Science(all)

Fingerprint

Dive into the research topics of 'Patchworking multiple pairwise distances for learning with distance matrices'. Together they form a unique fingerprint.

Cite this