Incremental commute time using random walks and online anomaly detection

Nguyen Lu Dang Khoa, Sanjay Chawla

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

1 Citation (Scopus)

Abstract

Commute time is a random walk based metric on graphs and has found widespread successful applications in many application domains. However, the computation of the commute time is expensive, involving the eigen decomposition of the graph Laplacian matrix. There has been effort to approximate the commute time in offline mode. Our interest is inspired by the use of commute time in online mode. We propose an accurate and efficient approximation for computing the commute time in an incremental fashion in order to facilitate real-time applications. An online anomaly detection technique is designed where the commute time of each new arriving data point to any data point in the current graph can be estimated in constant time ensuring a real-time response. The proposed approach shows its high accuracy and efficiency in many synthetic and real datasets and takes only 8 milliseconds on average to detect anomalies online on the DBLP graph which has more than 600,000 nodes and 2 millions edges.

Original languageEnglish
Title of host publicationMachine Learning and Knowledge Discovery in Databases - European Conference, ECML PKDD 2016, Proceedings
PublisherSpringer Verlag
Pages49-64
Number of pages16
Volume9851 LNAI
ISBN (Print)9783319461274
DOIs
Publication statusPublished - 2016
Event15th European Conference on Machine Learning and Principles and Practice of Knowledge Discovery in Databases, ECML PKDD 2016 - Riva del Garda, Italy
Duration: 19 Sep 201623 Sep 2016

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume9851 LNAI
ISSN (Print)03029743
ISSN (Electronic)16113349

Other

Other15th European Conference on Machine Learning and Principles and Practice of Knowledge Discovery in Databases, ECML PKDD 2016
CountryItaly
CityRiva del Garda
Period19/9/1623/9/16

Fingerprint

Anomaly Detection
Commute
Random walk
Graph in graph theory
Decomposition
Graph Laplacian
Real-time
Laplacian Matrix
Time Constant
Anomaly
High Efficiency
High Accuracy
Decompose
Metric
Computing
Approximation
Vertex of a graph

Keywords

  • Commute time
  • Incremental learning
  • Online anomaly detection
  • Random walk

ASJC Scopus subject areas

  • Theoretical Computer Science
  • Computer Science(all)

Cite this

Khoa, N. L. D., & Chawla, S. (2016). Incremental commute time using random walks and online anomaly detection. In Machine Learning and Knowledge Discovery in Databases - European Conference, ECML PKDD 2016, Proceedings (Vol. 9851 LNAI, pp. 49-64). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 9851 LNAI). Springer Verlag. https://doi.org/10.1007/978-3-319-46128-1_4

Incremental commute time using random walks and online anomaly detection. / Khoa, Nguyen Lu Dang; Chawla, Sanjay.

Machine Learning and Knowledge Discovery in Databases - European Conference, ECML PKDD 2016, Proceedings. Vol. 9851 LNAI Springer Verlag, 2016. p. 49-64 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 9851 LNAI).

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

Khoa, NLD & Chawla, S 2016, Incremental commute time using random walks and online anomaly detection. in Machine Learning and Knowledge Discovery in Databases - European Conference, ECML PKDD 2016, Proceedings. vol. 9851 LNAI, Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 9851 LNAI, Springer Verlag, pp. 49-64, 15th European Conference on Machine Learning and Principles and Practice of Knowledge Discovery in Databases, ECML PKDD 2016, Riva del Garda, Italy, 19/9/16. https://doi.org/10.1007/978-3-319-46128-1_4
Khoa NLD, Chawla S. Incremental commute time using random walks and online anomaly detection. In Machine Learning and Knowledge Discovery in Databases - European Conference, ECML PKDD 2016, Proceedings. Vol. 9851 LNAI. Springer Verlag. 2016. p. 49-64. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). https://doi.org/10.1007/978-3-319-46128-1_4
Khoa, Nguyen Lu Dang ; Chawla, Sanjay. / Incremental commute time using random walks and online anomaly detection. Machine Learning and Knowledge Discovery in Databases - European Conference, ECML PKDD 2016, Proceedings. Vol. 9851 LNAI Springer Verlag, 2016. pp. 49-64 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).
@inproceedings{4bb9f942ffac4bc6ba099e996d35a8b0,
title = "Incremental commute time using random walks and online anomaly detection",
abstract = "Commute time is a random walk based metric on graphs and has found widespread successful applications in many application domains. However, the computation of the commute time is expensive, involving the eigen decomposition of the graph Laplacian matrix. There has been effort to approximate the commute time in offline mode. Our interest is inspired by the use of commute time in online mode. We propose an accurate and efficient approximation for computing the commute time in an incremental fashion in order to facilitate real-time applications. An online anomaly detection technique is designed where the commute time of each new arriving data point to any data point in the current graph can be estimated in constant time ensuring a real-time response. The proposed approach shows its high accuracy and efficiency in many synthetic and real datasets and takes only 8 milliseconds on average to detect anomalies online on the DBLP graph which has more than 600,000 nodes and 2 millions edges.",
keywords = "Commute time, Incremental learning, Online anomaly detection, Random walk",
author = "Khoa, {Nguyen Lu Dang} and Sanjay Chawla",
year = "2016",
doi = "10.1007/978-3-319-46128-1_4",
language = "English",
isbn = "9783319461274",
volume = "9851 LNAI",
series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",
publisher = "Springer Verlag",
pages = "49--64",
booktitle = "Machine Learning and Knowledge Discovery in Databases - European Conference, ECML PKDD 2016, Proceedings",

}

TY - GEN

T1 - Incremental commute time using random walks and online anomaly detection

AU - Khoa, Nguyen Lu Dang

AU - Chawla, Sanjay

PY - 2016

Y1 - 2016

N2 - Commute time is a random walk based metric on graphs and has found widespread successful applications in many application domains. However, the computation of the commute time is expensive, involving the eigen decomposition of the graph Laplacian matrix. There has been effort to approximate the commute time in offline mode. Our interest is inspired by the use of commute time in online mode. We propose an accurate and efficient approximation for computing the commute time in an incremental fashion in order to facilitate real-time applications. An online anomaly detection technique is designed where the commute time of each new arriving data point to any data point in the current graph can be estimated in constant time ensuring a real-time response. The proposed approach shows its high accuracy and efficiency in many synthetic and real datasets and takes only 8 milliseconds on average to detect anomalies online on the DBLP graph which has more than 600,000 nodes and 2 millions edges.

AB - Commute time is a random walk based metric on graphs and has found widespread successful applications in many application domains. However, the computation of the commute time is expensive, involving the eigen decomposition of the graph Laplacian matrix. There has been effort to approximate the commute time in offline mode. Our interest is inspired by the use of commute time in online mode. We propose an accurate and efficient approximation for computing the commute time in an incremental fashion in order to facilitate real-time applications. An online anomaly detection technique is designed where the commute time of each new arriving data point to any data point in the current graph can be estimated in constant time ensuring a real-time response. The proposed approach shows its high accuracy and efficiency in many synthetic and real datasets and takes only 8 milliseconds on average to detect anomalies online on the DBLP graph which has more than 600,000 nodes and 2 millions edges.

KW - Commute time

KW - Incremental learning

KW - Online anomaly detection

KW - Random walk

UR - http://www.scopus.com/inward/record.url?scp=84988622219&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84988622219&partnerID=8YFLogxK

U2 - 10.1007/978-3-319-46128-1_4

DO - 10.1007/978-3-319-46128-1_4

M3 - Conference contribution

SN - 9783319461274

VL - 9851 LNAI

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 49

EP - 64

BT - Machine Learning and Knowledge Discovery in Databases - European Conference, ECML PKDD 2016, Proceedings

PB - Springer Verlag

ER -