Change-point detection with supervised learning and feature selection

Victor Eruhimov, Vladimir Martyanov, Eugene Tuv, George Runger

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

Abstract

Data streams with high dimensions are more and more common as data sets become wider. Time segments of stable system performance are often interrupted with change events. The change-point problem is to detect such changes and identify attributes that contribute to the change. Existing methods focus on detecting a single (or few) change-point in a univariate (or low-dimensional) process. We consider the important highdimensional multivariate case with multiple change-points and without an assumed distribution. The problem is transformed to a supervised learning problem with time as the output response and the process variables as inputs. This opens the problem to a wide set of supervised learning tools. Feature selection methods are used to identify the subset of variables that change. An illustrative example illustrates the method in an important type of application.

Original languageEnglish (US)
Title of host publicationICINCO 2007 - 4th International Conference on Informatics in Control, Automation and Robotics, Proceedings
Pages359-363
Number of pages5
VolumeICSO
StatePublished - 2007
Event4th International Conference on Informatics in Control, Automation and Robotics, ICINCO 2007 - Angers, France
Duration: May 9 2007May 12 2007

Other

Other4th International Conference on Informatics in Control, Automation and Robotics, ICINCO 2007
CountryFrance
CityAngers
Period5/9/075/12/07

Fingerprint

Supervised learning
Feature extraction

Keywords

  • Data streams
  • Ensembles
  • Multivariate control
  • Variable importance

ASJC Scopus subject areas

  • Artificial Intelligence
  • Computer Science Applications
  • Human-Computer Interaction
  • Software
  • Control and Systems Engineering
  • Electrical and Electronic Engineering

Cite this

Eruhimov, V., Martyanov, V., Tuv, E., & Runger, G. (2007). Change-point detection with supervised learning and feature selection. In ICINCO 2007 - 4th International Conference on Informatics in Control, Automation and Robotics, Proceedings (Vol. ICSO, pp. 359-363)

Change-point detection with supervised learning and feature selection. / Eruhimov, Victor; Martyanov, Vladimir; Tuv, Eugene; Runger, George.

ICINCO 2007 - 4th International Conference on Informatics in Control, Automation and Robotics, Proceedings. Vol. ICSO 2007. p. 359-363.

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

Eruhimov, V, Martyanov, V, Tuv, E & Runger, G 2007, Change-point detection with supervised learning and feature selection. in ICINCO 2007 - 4th International Conference on Informatics in Control, Automation and Robotics, Proceedings. vol. ICSO, pp. 359-363, 4th International Conference on Informatics in Control, Automation and Robotics, ICINCO 2007, Angers, France, 5/9/07.
Eruhimov V, Martyanov V, Tuv E, Runger G. Change-point detection with supervised learning and feature selection. In ICINCO 2007 - 4th International Conference on Informatics in Control, Automation and Robotics, Proceedings. Vol. ICSO. 2007. p. 359-363
Eruhimov, Victor ; Martyanov, Vladimir ; Tuv, Eugene ; Runger, George. / Change-point detection with supervised learning and feature selection. ICINCO 2007 - 4th International Conference on Informatics in Control, Automation and Robotics, Proceedings. Vol. ICSO 2007. pp. 359-363
@inproceedings{0be769185c21469fa7025deef3ae81f6,
title = "Change-point detection with supervised learning and feature selection",
abstract = "Data streams with high dimensions are more and more common as data sets become wider. Time segments of stable system performance are often interrupted with change events. The change-point problem is to detect such changes and identify attributes that contribute to the change. Existing methods focus on detecting a single (or few) change-point in a univariate (or low-dimensional) process. We consider the important highdimensional multivariate case with multiple change-points and without an assumed distribution. The problem is transformed to a supervised learning problem with time as the output response and the process variables as inputs. This opens the problem to a wide set of supervised learning tools. Feature selection methods are used to identify the subset of variables that change. An illustrative example illustrates the method in an important type of application.",
keywords = "Data streams, Ensembles, Multivariate control, Variable importance",
author = "Victor Eruhimov and Vladimir Martyanov and Eugene Tuv and George Runger",
year = "2007",
language = "English (US)",
volume = "ICSO",
pages = "359--363",
booktitle = "ICINCO 2007 - 4th International Conference on Informatics in Control, Automation and Robotics, Proceedings",

}

TY - GEN

T1 - Change-point detection with supervised learning and feature selection

AU - Eruhimov, Victor

AU - Martyanov, Vladimir

AU - Tuv, Eugene

AU - Runger, George

PY - 2007

Y1 - 2007

N2 - Data streams with high dimensions are more and more common as data sets become wider. Time segments of stable system performance are often interrupted with change events. The change-point problem is to detect such changes and identify attributes that contribute to the change. Existing methods focus on detecting a single (or few) change-point in a univariate (or low-dimensional) process. We consider the important highdimensional multivariate case with multiple change-points and without an assumed distribution. The problem is transformed to a supervised learning problem with time as the output response and the process variables as inputs. This opens the problem to a wide set of supervised learning tools. Feature selection methods are used to identify the subset of variables that change. An illustrative example illustrates the method in an important type of application.

AB - Data streams with high dimensions are more and more common as data sets become wider. Time segments of stable system performance are often interrupted with change events. The change-point problem is to detect such changes and identify attributes that contribute to the change. Existing methods focus on detecting a single (or few) change-point in a univariate (or low-dimensional) process. We consider the important highdimensional multivariate case with multiple change-points and without an assumed distribution. The problem is transformed to a supervised learning problem with time as the output response and the process variables as inputs. This opens the problem to a wide set of supervised learning tools. Feature selection methods are used to identify the subset of variables that change. An illustrative example illustrates the method in an important type of application.

KW - Data streams

KW - Ensembles

KW - Multivariate control

KW - Variable importance

UR - http://www.scopus.com/inward/record.url?scp=67149136281&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=67149136281&partnerID=8YFLogxK

M3 - Conference contribution

AN - SCOPUS:67149136281

VL - ICSO

SP - 359

EP - 363

BT - ICINCO 2007 - 4th International Conference on Informatics in Control, Automation and Robotics, Proceedings

ER -