Detection of unique temporal segments by information theoretic meta-clustering

Shin Ando, Einoshin Suzuki

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

4 Citations (Scopus)

Abstract

The central challenge in temporal data analysis is to obtain knowledge about its underlying dynamics. In this paper, we address the observation of noisy, stochastic processes and attempt to detect temporal segments that are related to inconsistencies and irregularities in its dynamics. Many conventional anomaly detection approaches detect anomalies based on the distance between patterns, and often provide only limited intuition about the generative process of the anomalies. Meanwhile, model-based approaches have difficulty in identifying a small, clustered set of anomalies. We propose Information- theoretic Meta-clustering (ITMC), a formalization of model-based clustering principled by the theory of lossy data compression. ITMC identifies a 'unique' cluster whose distribution diverges significantly from the entire dataset. Furthermore, ITMC employs a regularization term derived from the preference for high compression rate, which is critical to the precision of detection. For empirical evaluation, we apply ITMC to two temporal anomaly detection tasks. Datasets are taken from generative processes involving heterogeneous and inconsistent dynamics. A comparison to baseline methods shows that the proposed algorithm detects segments from irregular states with significantly high precision and recall.

Original languageEnglish
Title of host publicationKDD '09
Subtitle of host publicationProceedings of the 15th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining
Pages59-67
Number of pages9
DOIs
Publication statusPublished - Nov 9 2009
Event15th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, KDD '09 - Paris, France
Duration: Jun 28 2009Jul 1 2009

Publication series

NameProceedings of the ACM SIGKDD International Conference on Knowledge Discovery and Data Mining

Other

Other15th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, KDD '09
CountryFrance
CityParis
Period6/28/097/1/09

Fingerprint

Data compression
Random processes

All Science Journal Classification (ASJC) codes

  • Software
  • Information Systems

Cite this

Ando, S., & Suzuki, E. (2009). Detection of unique temporal segments by information theoretic meta-clustering. In KDD '09: Proceedings of the 15th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining (pp. 59-67). (Proceedings of the ACM SIGKDD International Conference on Knowledge Discovery and Data Mining). https://doi.org/10.1145/1557019.1557033

Detection of unique temporal segments by information theoretic meta-clustering. / Ando, Shin; Suzuki, Einoshin.

KDD '09: Proceedings of the 15th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining. 2009. p. 59-67 (Proceedings of the ACM SIGKDD International Conference on Knowledge Discovery and Data Mining).

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

Ando, S & Suzuki, E 2009, Detection of unique temporal segments by information theoretic meta-clustering. in KDD '09: Proceedings of the 15th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining. Proceedings of the ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, pp. 59-67, 15th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, KDD '09, Paris, France, 6/28/09. https://doi.org/10.1145/1557019.1557033
Ando S, Suzuki E. Detection of unique temporal segments by information theoretic meta-clustering. In KDD '09: Proceedings of the 15th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining. 2009. p. 59-67. (Proceedings of the ACM SIGKDD International Conference on Knowledge Discovery and Data Mining). https://doi.org/10.1145/1557019.1557033
Ando, Shin ; Suzuki, Einoshin. / Detection of unique temporal segments by information theoretic meta-clustering. KDD '09: Proceedings of the 15th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining. 2009. pp. 59-67 (Proceedings of the ACM SIGKDD International Conference on Knowledge Discovery and Data Mining).
@inproceedings{6efa7e7a4b154219b0754c5e7f48fdcb,
title = "Detection of unique temporal segments by information theoretic meta-clustering",
abstract = "The central challenge in temporal data analysis is to obtain knowledge about its underlying dynamics. In this paper, we address the observation of noisy, stochastic processes and attempt to detect temporal segments that are related to inconsistencies and irregularities in its dynamics. Many conventional anomaly detection approaches detect anomalies based on the distance between patterns, and often provide only limited intuition about the generative process of the anomalies. Meanwhile, model-based approaches have difficulty in identifying a small, clustered set of anomalies. We propose Information- theoretic Meta-clustering (ITMC), a formalization of model-based clustering principled by the theory of lossy data compression. ITMC identifies a 'unique' cluster whose distribution diverges significantly from the entire dataset. Furthermore, ITMC employs a regularization term derived from the preference for high compression rate, which is critical to the precision of detection. For empirical evaluation, we apply ITMC to two temporal anomaly detection tasks. Datasets are taken from generative processes involving heterogeneous and inconsistent dynamics. A comparison to baseline methods shows that the proposed algorithm detects segments from irregular states with significantly high precision and recall.",
author = "Shin Ando and Einoshin Suzuki",
year = "2009",
month = "11",
day = "9",
doi = "10.1145/1557019.1557033",
language = "English",
isbn = "9781605584959",
series = "Proceedings of the ACM SIGKDD International Conference on Knowledge Discovery and Data Mining",
pages = "59--67",
booktitle = "KDD '09",

}

TY - GEN

T1 - Detection of unique temporal segments by information theoretic meta-clustering

AU - Ando, Shin

AU - Suzuki, Einoshin

PY - 2009/11/9

Y1 - 2009/11/9

N2 - The central challenge in temporal data analysis is to obtain knowledge about its underlying dynamics. In this paper, we address the observation of noisy, stochastic processes and attempt to detect temporal segments that are related to inconsistencies and irregularities in its dynamics. Many conventional anomaly detection approaches detect anomalies based on the distance between patterns, and often provide only limited intuition about the generative process of the anomalies. Meanwhile, model-based approaches have difficulty in identifying a small, clustered set of anomalies. We propose Information- theoretic Meta-clustering (ITMC), a formalization of model-based clustering principled by the theory of lossy data compression. ITMC identifies a 'unique' cluster whose distribution diverges significantly from the entire dataset. Furthermore, ITMC employs a regularization term derived from the preference for high compression rate, which is critical to the precision of detection. For empirical evaluation, we apply ITMC to two temporal anomaly detection tasks. Datasets are taken from generative processes involving heterogeneous and inconsistent dynamics. A comparison to baseline methods shows that the proposed algorithm detects segments from irregular states with significantly high precision and recall.

AB - The central challenge in temporal data analysis is to obtain knowledge about its underlying dynamics. In this paper, we address the observation of noisy, stochastic processes and attempt to detect temporal segments that are related to inconsistencies and irregularities in its dynamics. Many conventional anomaly detection approaches detect anomalies based on the distance between patterns, and often provide only limited intuition about the generative process of the anomalies. Meanwhile, model-based approaches have difficulty in identifying a small, clustered set of anomalies. We propose Information- theoretic Meta-clustering (ITMC), a formalization of model-based clustering principled by the theory of lossy data compression. ITMC identifies a 'unique' cluster whose distribution diverges significantly from the entire dataset. Furthermore, ITMC employs a regularization term derived from the preference for high compression rate, which is critical to the precision of detection. For empirical evaluation, we apply ITMC to two temporal anomaly detection tasks. Datasets are taken from generative processes involving heterogeneous and inconsistent dynamics. A comparison to baseline methods shows that the proposed algorithm detects segments from irregular states with significantly high precision and recall.

UR - http://www.scopus.com/inward/record.url?scp=70350668614&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=70350668614&partnerID=8YFLogxK

U2 - 10.1145/1557019.1557033

DO - 10.1145/1557019.1557033

M3 - Conference contribution

AN - SCOPUS:70350668614

SN - 9781605584959

T3 - Proceedings of the ACM SIGKDD International Conference on Knowledge Discovery and Data Mining

SP - 59

EP - 67

BT - KDD '09

ER -