Subspace Determination Through Local Intrinsic Dimensional Decomposition

Ruben Becker, Imane Hafnaoui, Michael E. Houle, Pan Li, Arthur Zimek

Research output: Chapter in Book/Report/Conference proceedingArticle in proceedingsResearchpeer-review

Abstract

Axis-aligned subspace clustering generally entails searching through enormous numbers of subspaces (feature combinations) and evaluation of cluster quality within each subspace. In this paper, we tackle the problem of identifying subsets of features with the most significant contribution to the formation of the local neighborhood surrounding a given data point. For each point, the recently-proposed Local Intrinsic Dimension (LID) model is used in identifying the axis directions along which features have the greatest local discriminability, or equivalently, the fewest number of components of LID that capture the local complexity of the data. In this paper, we develop an estimator of LID along axis projections, and provide preliminary evidence that this LID decomposition can indicate axis-aligned data subspaces that support the formation of clusters.
Original languageEnglish
Title of host publicationSimilarity Search and Applications - 12th International Conference, SISAP 2019, Newark, NJ, USA, October 2-4, 2019, Proceedings
EditorsGiuseppe Amato, Claudio Gennaro, Vincent Oria, Miloš Radovanovic
PublisherSpringer
Publication date2019
Pages281-289
ISBN (Print)978-3-030-32046-1
ISBN (Electronic)978-3-030-32047-8
DOIs
Publication statusPublished - 2019
EventInternational Conference on Similarity Search and Applications - New Jersey Institute of Technology (NJIT), Newark, United States
Duration: 2. Oct 20194. Oct 2019
Conference number: 12
http://www.sisap.org/2019/

Conference

ConferenceInternational Conference on Similarity Search and Applications
Number12
LocationNew Jersey Institute of Technology (NJIT)
CountryUnited States
CityNewark
Period02/10/201904/10/2019
Internet address
SeriesLecture Notes in Computer Science
Volume11807
ISSN0302-9743

Fingerprint

Decomposition

Keywords

  • Estimation
  • Intrinsic dimensionality
  • Subspace

Cite this

Becker, R., Hafnaoui, I., Houle, M. E., Li, P., & Zimek, A. (2019). Subspace Determination Through Local Intrinsic Dimensional Decomposition. In G. Amato, C. Gennaro, V. Oria, & M. Radovanovic (Eds.), Similarity Search and Applications - 12th International Conference, SISAP 2019, Newark, NJ, USA, October 2-4, 2019, Proceedings (pp. 281-289). Springer. Lecture Notes in Computer Science, Vol.. 11807 https://doi.org/10.1007/978-3-030-32047-8_25
Becker, Ruben ; Hafnaoui, Imane ; Houle, Michael E. ; Li, Pan ; Zimek, Arthur. / Subspace Determination Through Local Intrinsic Dimensional Decomposition. Similarity Search and Applications - 12th International Conference, SISAP 2019, Newark, NJ, USA, October 2-4, 2019, Proceedings. editor / Giuseppe Amato ; Claudio Gennaro ; Vincent Oria ; Miloš Radovanovic. Springer, 2019. pp. 281-289 (Lecture Notes in Computer Science, Vol. 11807).
@inproceedings{9c756bc590f845a9bfc585a1635b3b08,
title = "Subspace Determination Through Local Intrinsic Dimensional Decomposition",
abstract = "Axis-aligned subspace clustering generally entails searching through enormous numbers of subspaces (feature combinations) and evaluation of cluster quality within each subspace. In this paper, we tackle the problem of identifying subsets of features with the most significant contribution to the formation of the local neighborhood surrounding a given data point. For each point, the recently-proposed Local Intrinsic Dimension (LID) model is used in identifying the axis directions along which features have the greatest local discriminability, or equivalently, the fewest number of components of LID that capture the local complexity of the data. In this paper, we develop an estimator of LID along axis projections, and provide preliminary evidence that this LID decomposition can indicate axis-aligned data subspaces that support the formation of clusters.",
keywords = "Estimation, Intrinsic dimensionality, Subspace",
author = "Ruben Becker and Imane Hafnaoui and Houle, {Michael E.} and Pan Li and Arthur Zimek",
year = "2019",
doi = "10.1007/978-3-030-32047-8_25",
language = "English",
isbn = "978-3-030-32046-1",
series = "Lecture Notes in Computer Science",
publisher = "Springer",
pages = "281--289",
editor = "Giuseppe Amato and Claudio Gennaro and Vincent Oria and Miloš Radovanovic",
booktitle = "Similarity Search and Applications - 12th International Conference, SISAP 2019, Newark, NJ, USA, October 2-4, 2019, Proceedings",
address = "Germany",

}

Becker, R, Hafnaoui, I, Houle, ME, Li, P & Zimek, A 2019, Subspace Determination Through Local Intrinsic Dimensional Decomposition. in G Amato, C Gennaro, V Oria & M Radovanovic (eds), Similarity Search and Applications - 12th International Conference, SISAP 2019, Newark, NJ, USA, October 2-4, 2019, Proceedings. Springer, Lecture Notes in Computer Science, vol. 11807, pp. 281-289, International Conference on Similarity Search and Applications, Newark, United States, 02/10/2019. https://doi.org/10.1007/978-3-030-32047-8_25

Subspace Determination Through Local Intrinsic Dimensional Decomposition. / Becker, Ruben; Hafnaoui, Imane; Houle, Michael E.; Li, Pan; Zimek, Arthur.

Similarity Search and Applications - 12th International Conference, SISAP 2019, Newark, NJ, USA, October 2-4, 2019, Proceedings. ed. / Giuseppe Amato; Claudio Gennaro; Vincent Oria; Miloš Radovanovic. Springer, 2019. p. 281-289 (Lecture Notes in Computer Science, Vol. 11807).

Research output: Chapter in Book/Report/Conference proceedingArticle in proceedingsResearchpeer-review

TY - GEN

T1 - Subspace Determination Through Local Intrinsic Dimensional Decomposition

AU - Becker, Ruben

AU - Hafnaoui, Imane

AU - Houle, Michael E.

AU - Li, Pan

AU - Zimek, Arthur

PY - 2019

Y1 - 2019

N2 - Axis-aligned subspace clustering generally entails searching through enormous numbers of subspaces (feature combinations) and evaluation of cluster quality within each subspace. In this paper, we tackle the problem of identifying subsets of features with the most significant contribution to the formation of the local neighborhood surrounding a given data point. For each point, the recently-proposed Local Intrinsic Dimension (LID) model is used in identifying the axis directions along which features have the greatest local discriminability, or equivalently, the fewest number of components of LID that capture the local complexity of the data. In this paper, we develop an estimator of LID along axis projections, and provide preliminary evidence that this LID decomposition can indicate axis-aligned data subspaces that support the formation of clusters.

AB - Axis-aligned subspace clustering generally entails searching through enormous numbers of subspaces (feature combinations) and evaluation of cluster quality within each subspace. In this paper, we tackle the problem of identifying subsets of features with the most significant contribution to the formation of the local neighborhood surrounding a given data point. For each point, the recently-proposed Local Intrinsic Dimension (LID) model is used in identifying the axis directions along which features have the greatest local discriminability, or equivalently, the fewest number of components of LID that capture the local complexity of the data. In this paper, we develop an estimator of LID along axis projections, and provide preliminary evidence that this LID decomposition can indicate axis-aligned data subspaces that support the formation of clusters.

KW - Estimation

KW - Intrinsic dimensionality

KW - Subspace

U2 - 10.1007/978-3-030-32047-8_25

DO - 10.1007/978-3-030-32047-8_25

M3 - Article in proceedings

SN - 978-3-030-32046-1

T3 - Lecture Notes in Computer Science

SP - 281

EP - 289

BT - Similarity Search and Applications - 12th International Conference, SISAP 2019, Newark, NJ, USA, October 2-4, 2019, Proceedings

A2 - Amato, Giuseppe

A2 - Gennaro, Claudio

A2 - Oria, Vincent

A2 - Radovanovic, Miloš

PB - Springer

ER -

Becker R, Hafnaoui I, Houle ME, Li P, Zimek A. Subspace Determination Through Local Intrinsic Dimensional Decomposition. In Amato G, Gennaro C, Oria V, Radovanovic M, editors, Similarity Search and Applications - 12th International Conference, SISAP 2019, Newark, NJ, USA, October 2-4, 2019, Proceedings. Springer. 2019. p. 281-289. (Lecture Notes in Computer Science, Vol. 11807). https://doi.org/10.1007/978-3-030-32047-8_25