Abstract

Anomaly detection in data streams requires a signal of an unusual event, and an actionable response requires diagnostics. Furthermore, monitoring for process control is often concerned with one or more target (controlled) attributes. Consequently, it is necessary to separate anomalies (and their contributing attributes) that could influence the controlled target strongly, and this becomes more important with the increased number of monitored attributes in modern processes. This task leads to a difficult problem not addressed directly by the machine learning/process control community. We introduce the target-aware anomaly detection problem and present a solution for process control in modern systems (with nonlinear dependencies, high dimensional noisy data, missing data, and so on). The main objective is to identify and rank outliers and also diagnose their contributing attributes with respect to the possible effect on the response. The method is different from traditional linear and/or univariate approaches, as it can deal with local data structure in the neighborhood of an outlier, and can handle complex interactions via the use of an appropriate learner. In addition, the method can be computed quickly and does not require time consuming matrix operations. Comparisons are made to traditional contribution plots computed from partial least squares.

Original languageEnglish (US)
Title of host publicationICINCO 2011 - Proceedings of the 8th International Conference on Informatics in Control, Automation and Robotics
Pages14-23
Number of pages10
Volume1
StatePublished - 2011
Event8th International Conference on Informatics in Control, Automation and Robotics, ICINCO 2011 - Noordwijkerhout, Netherlands
Duration: Jul 28 2011Jul 31 2011

Other

Other8th International Conference on Informatics in Control, Automation and Robotics, ICINCO 2011
CountryNetherlands
CityNoordwijkerhout
Period7/28/117/31/11

Fingerprint

Process control
Data structures
Learning systems
Monitoring

Keywords

  • Contribution plots
  • Outliers
  • Partial least squares
  • Process control

ASJC Scopus subject areas

  • Information Systems
  • Control and Systems Engineering

Cite this

Borisov, A., Runger, G., & Tuv, E. (2011). Target-aware anomaly detection and diagnosis. In ICINCO 2011 - Proceedings of the 8th International Conference on Informatics in Control, Automation and Robotics (Vol. 1, pp. 14-23)

Target-aware anomaly detection and diagnosis. / Borisov, Alexander; Runger, George; Tuv, Eugene.

ICINCO 2011 - Proceedings of the 8th International Conference on Informatics in Control, Automation and Robotics. Vol. 1 2011. p. 14-23.

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

Borisov, A, Runger, G & Tuv, E 2011, Target-aware anomaly detection and diagnosis. in ICINCO 2011 - Proceedings of the 8th International Conference on Informatics in Control, Automation and Robotics. vol. 1, pp. 14-23, 8th International Conference on Informatics in Control, Automation and Robotics, ICINCO 2011, Noordwijkerhout, Netherlands, 7/28/11.
Borisov A, Runger G, Tuv E. Target-aware anomaly detection and diagnosis. In ICINCO 2011 - Proceedings of the 8th International Conference on Informatics in Control, Automation and Robotics. Vol. 1. 2011. p. 14-23
Borisov, Alexander ; Runger, George ; Tuv, Eugene. / Target-aware anomaly detection and diagnosis. ICINCO 2011 - Proceedings of the 8th International Conference on Informatics in Control, Automation and Robotics. Vol. 1 2011. pp. 14-23
@inproceedings{db8c13f497384b508554ac8c533e1f8f,
title = "Target-aware anomaly detection and diagnosis",
abstract = "Anomaly detection in data streams requires a signal of an unusual event, and an actionable response requires diagnostics. Furthermore, monitoring for process control is often concerned with one or more target (controlled) attributes. Consequently, it is necessary to separate anomalies (and their contributing attributes) that could influence the controlled target strongly, and this becomes more important with the increased number of monitored attributes in modern processes. This task leads to a difficult problem not addressed directly by the machine learning/process control community. We introduce the target-aware anomaly detection problem and present a solution for process control in modern systems (with nonlinear dependencies, high dimensional noisy data, missing data, and so on). The main objective is to identify and rank outliers and also diagnose their contributing attributes with respect to the possible effect on the response. The method is different from traditional linear and/or univariate approaches, as it can deal with local data structure in the neighborhood of an outlier, and can handle complex interactions via the use of an appropriate learner. In addition, the method can be computed quickly and does not require time consuming matrix operations. Comparisons are made to traditional contribution plots computed from partial least squares.",
keywords = "Contribution plots, Outliers, Partial least squares, Process control",
author = "Alexander Borisov and George Runger and Eugene Tuv",
year = "2011",
language = "English (US)",
isbn = "9789898425744",
volume = "1",
pages = "14--23",
booktitle = "ICINCO 2011 - Proceedings of the 8th International Conference on Informatics in Control, Automation and Robotics",

}

TY - GEN

T1 - Target-aware anomaly detection and diagnosis

AU - Borisov, Alexander

AU - Runger, George

AU - Tuv, Eugene

PY - 2011

Y1 - 2011

N2 - Anomaly detection in data streams requires a signal of an unusual event, and an actionable response requires diagnostics. Furthermore, monitoring for process control is often concerned with one or more target (controlled) attributes. Consequently, it is necessary to separate anomalies (and their contributing attributes) that could influence the controlled target strongly, and this becomes more important with the increased number of monitored attributes in modern processes. This task leads to a difficult problem not addressed directly by the machine learning/process control community. We introduce the target-aware anomaly detection problem and present a solution for process control in modern systems (with nonlinear dependencies, high dimensional noisy data, missing data, and so on). The main objective is to identify and rank outliers and also diagnose their contributing attributes with respect to the possible effect on the response. The method is different from traditional linear and/or univariate approaches, as it can deal with local data structure in the neighborhood of an outlier, and can handle complex interactions via the use of an appropriate learner. In addition, the method can be computed quickly and does not require time consuming matrix operations. Comparisons are made to traditional contribution plots computed from partial least squares.

AB - Anomaly detection in data streams requires a signal of an unusual event, and an actionable response requires diagnostics. Furthermore, monitoring for process control is often concerned with one or more target (controlled) attributes. Consequently, it is necessary to separate anomalies (and their contributing attributes) that could influence the controlled target strongly, and this becomes more important with the increased number of monitored attributes in modern processes. This task leads to a difficult problem not addressed directly by the machine learning/process control community. We introduce the target-aware anomaly detection problem and present a solution for process control in modern systems (with nonlinear dependencies, high dimensional noisy data, missing data, and so on). The main objective is to identify and rank outliers and also diagnose their contributing attributes with respect to the possible effect on the response. The method is different from traditional linear and/or univariate approaches, as it can deal with local data structure in the neighborhood of an outlier, and can handle complex interactions via the use of an appropriate learner. In addition, the method can be computed quickly and does not require time consuming matrix operations. Comparisons are made to traditional contribution plots computed from partial least squares.

KW - Contribution plots

KW - Outliers

KW - Partial least squares

KW - Process control

UR - http://www.scopus.com/inward/record.url?scp=80052568128&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=80052568128&partnerID=8YFLogxK

M3 - Conference contribution

AN - SCOPUS:80052568128

SN - 9789898425744

VL - 1

SP - 14

EP - 23

BT - ICINCO 2011 - Proceedings of the 8th International Conference on Informatics in Control, Automation and Robotics

ER -