### 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 language | English |
---|---|

Title of host publication | Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) |

Publisher | Springer Verlag |

Pages | 287-294 |

Number of pages | 8 |

Volume | 9237 |

ISBN (Print) | 9783319224817 |

DOIs | |

Publication status | Published - 2015 |

Event | 12th International Conference on Latent Variable Analysis and Signal Separation, LVA/ICA 2015 - Liberec, Czech Republic Duration: 2015 Aug 25 → 2015 Aug 28 |

### Publication series

Name | Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) |
---|---|

Volume | 9237 |

ISSN (Print) | 03029743 |

ISSN (Electronic) | 16113349 |

### Other

Other | 12th International Conference on Latent Variable Analysis and Signal Separation, LVA/ICA 2015 |
---|---|

Country | Czech Republic |

City | Liberec |

Period | 15/8/25 → 15/8/28 |

### Fingerprint

### Keywords

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

### ASJC Scopus subject areas

- Computer Science(all)
- Theoretical Computer Science

### Cite this

*Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)*(Vol. 9237, pp. 287-294). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 9237). Springer Verlag. https://doi.org/10.1007/978-3-319-22482-4_33

**Patchworking multiple pairwise distances for learning with distance matrices.** / Takano, Ken; Hino, Hideitsu; Yoshikawa, Yuki; Murata, Noboru.

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

*Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics).*vol. 9237, Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 9237, Springer Verlag, pp. 287-294, 12th International Conference on Latent Variable Analysis and Signal Separation, LVA/ICA 2015, Liberec, Czech Republic, 15/8/25. https://doi.org/10.1007/978-3-319-22482-4_33

}

TY - GEN

T1 - Patchworking multiple pairwise distances for learning with distance matrices

AU - Takano, Ken

AU - Hino, Hideitsu

AU - Yoshikawa, Yuki

AU - Murata, Noboru

PY - 2015

Y1 - 2015

N2 - 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.

AB - 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.

KW - Classification

KW - Decision trees

KW - Graph kernel

KW - Random forest

KW - Spike train

KW - Structured data

UR - http://www.scopus.com/inward/record.url?scp=84944681542&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84944681542&partnerID=8YFLogxK

U2 - 10.1007/978-3-319-22482-4_33

DO - 10.1007/978-3-319-22482-4_33

M3 - Conference contribution

AN - SCOPUS:84944681542

SN - 9783319224817

VL - 9237

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 287

EP - 294

BT - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

PB - Springer Verlag

ER -