Modelling recurrent events for improving online change detection

A. Maslov, M. Pechenizkiy, I. Žliobaite, T. Kärkkäinen

Research output: Chapter in Book/Report/Conference proceedingConference contributionAcademicpeer-review

2 Citations (Scopus)

Abstract


The task of online change point detection in sensor data streams is often complicated due to presence of noise that can be mistaken for real changes and therefore affecting performance of change detectors. Most of the existing change detection methods assume that changes are independent from each other and occur at random in time. In this paper we study how performance of detectors can be improved in case of recurrent changes. We analytically demonstrate under which conditions and for how long recurrence information is useful for improving the detection accuracy. We propose a simple computationally efficient message passing procedure for calculating a predictive probability distribution of change occurrence in the future. We demonstrate two straightforward ways to apply the proposed procedure to existing change detection algorithms. Our experimental analysis illustrates the effectiveness of these approaches in improving the performance of a baseline online change detector by incorporating recurrence information.


Read More: http://epubs.siam.org/doi/10.1137/1.9781611974348.62
Original languageEnglish
Title of host publication2016 SIAM International Conference on Data Mining (SDM16), 5-7 May 2016, Maimi, Florida
Place of Publications.l.
PublisherSociety for Industrial and Applied Mathematics (SIAM)
Pages549-557
DOIs
Publication statusPublished - 2016

Fingerprint

Detectors
Message passing
Probability distributions
Sensors

Cite this

Maslov, A., Pechenizkiy, M., Žliobaite, I., & Kärkkäinen, T. (2016). Modelling recurrent events for improving online change detection. In 2016 SIAM International Conference on Data Mining (SDM16), 5-7 May 2016, Maimi, Florida (pp. 549-557). s.l.: Society for Industrial and Applied Mathematics (SIAM). https://doi.org/10.1137/1.9781611974348.62
Maslov, A. ; Pechenizkiy, M. ; Žliobaite, I. ; Kärkkäinen, T. / Modelling recurrent events for improving online change detection. 2016 SIAM International Conference on Data Mining (SDM16), 5-7 May 2016, Maimi, Florida. s.l. : Society for Industrial and Applied Mathematics (SIAM), 2016. pp. 549-557
@inproceedings{002157bdfcdb43939d6a75f0ac0aa48d,
title = "Modelling recurrent events for improving online change detection",
abstract = "The task of online change point detection in sensor data streams is often complicated due to presence of noise that can be mistaken for real changes and therefore affecting performance of change detectors. Most of the existing change detection methods assume that changes are independent from each other and occur at random in time. In this paper we study how performance of detectors can be improved in case of recurrent changes. We analytically demonstrate under which conditions and for how long recurrence information is useful for improving the detection accuracy. We propose a simple computationally efficient message passing procedure for calculating a predictive probability distribution of change occurrence in the future. We demonstrate two straightforward ways to apply the proposed procedure to existing change detection algorithms. Our experimental analysis illustrates the effectiveness of these approaches in improving the performance of a baseline online change detector by incorporating recurrence information.Read More: http://epubs.siam.org/doi/10.1137/1.9781611974348.62",
author = "A. Maslov and M. Pechenizkiy and I. Žliobaite and T. K{\"a}rkk{\"a}inen",
year = "2016",
doi = "10.1137/1.9781611974348.62",
language = "English",
pages = "549--557",
booktitle = "2016 SIAM International Conference on Data Mining (SDM16), 5-7 May 2016, Maimi, Florida",
publisher = "Society for Industrial and Applied Mathematics (SIAM)",
address = "United States",

}

Maslov, A, Pechenizkiy, M, Žliobaite, I & Kärkkäinen, T 2016, Modelling recurrent events for improving online change detection. in 2016 SIAM International Conference on Data Mining (SDM16), 5-7 May 2016, Maimi, Florida. Society for Industrial and Applied Mathematics (SIAM), s.l., pp. 549-557. https://doi.org/10.1137/1.9781611974348.62

Modelling recurrent events for improving online change detection. / Maslov, A.; Pechenizkiy, M.; Žliobaite, I.; Kärkkäinen, T.

2016 SIAM International Conference on Data Mining (SDM16), 5-7 May 2016, Maimi, Florida. s.l. : Society for Industrial and Applied Mathematics (SIAM), 2016. p. 549-557.

Research output: Chapter in Book/Report/Conference proceedingConference contributionAcademicpeer-review

TY - GEN

T1 - Modelling recurrent events for improving online change detection

AU - Maslov, A.

AU - Pechenizkiy, M.

AU - Žliobaite, I.

AU - Kärkkäinen, T.

PY - 2016

Y1 - 2016

N2 - The task of online change point detection in sensor data streams is often complicated due to presence of noise that can be mistaken for real changes and therefore affecting performance of change detectors. Most of the existing change detection methods assume that changes are independent from each other and occur at random in time. In this paper we study how performance of detectors can be improved in case of recurrent changes. We analytically demonstrate under which conditions and for how long recurrence information is useful for improving the detection accuracy. We propose a simple computationally efficient message passing procedure for calculating a predictive probability distribution of change occurrence in the future. We demonstrate two straightforward ways to apply the proposed procedure to existing change detection algorithms. Our experimental analysis illustrates the effectiveness of these approaches in improving the performance of a baseline online change detector by incorporating recurrence information.Read More: http://epubs.siam.org/doi/10.1137/1.9781611974348.62

AB - The task of online change point detection in sensor data streams is often complicated due to presence of noise that can be mistaken for real changes and therefore affecting performance of change detectors. Most of the existing change detection methods assume that changes are independent from each other and occur at random in time. In this paper we study how performance of detectors can be improved in case of recurrent changes. We analytically demonstrate under which conditions and for how long recurrence information is useful for improving the detection accuracy. We propose a simple computationally efficient message passing procedure for calculating a predictive probability distribution of change occurrence in the future. We demonstrate two straightforward ways to apply the proposed procedure to existing change detection algorithms. Our experimental analysis illustrates the effectiveness of these approaches in improving the performance of a baseline online change detector by incorporating recurrence information.Read More: http://epubs.siam.org/doi/10.1137/1.9781611974348.62

U2 - 10.1137/1.9781611974348.62

DO - 10.1137/1.9781611974348.62

M3 - Conference contribution

SP - 549

EP - 557

BT - 2016 SIAM International Conference on Data Mining (SDM16), 5-7 May 2016, Maimi, Florida

PB - Society for Industrial and Applied Mathematics (SIAM)

CY - s.l.

ER -

Maslov A, Pechenizkiy M, Žliobaite I, Kärkkäinen T. Modelling recurrent events for improving online change detection. In 2016 SIAM International Conference on Data Mining (SDM16), 5-7 May 2016, Maimi, Florida. s.l.: Society for Industrial and Applied Mathematics (SIAM). 2016. p. 549-557 https://doi.org/10.1137/1.9781611974348.62