Incremental commute time and its online applications

Nguyen Lu Dang Khoa, Yang Wang, Sanjay Chawla

Research output: Contribution to journalArticle

Abstract

Commute time is a robust measure on graphs based on random walks. It has been successfully applied in many application domains including personalized search, collaborative filtering and network intrusion detection. However, the computation of the commute time is expensive since it involves the eigen decomposition of the graph Laplacian matrix. There has been effort to approximate the commute time but they only work in an offline mode. In this work, an accurate and efficient approximation for computing the commute time is proposed in an incremental fashion in order to facilitate online applications. Using the incremental commutime, we design an online anomaly detection application where the commute time of each new arriving data point to any point in the current graph can be estimated in constant time. The proposed approach shows its high accuracy and efficiency in synthetic and real datasets for online applications. It 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. We also discuss the use of incremental commute time for other online applications such as classification, graph ranking and clustering.

Original languageEnglish
Pages (from-to)101-112
Number of pages12
JournalPattern Recognition
Volume88
DOIs
Publication statusPublished - 1 Apr 2019

Fingerprint

Collaborative filtering
Intrusion detection
Decomposition

Keywords

  • Anomaly detection
  • Commute time
  • Manifold learning
  • Online learning
  • Random walks

ASJC Scopus subject areas

  • Software
  • Signal Processing
  • Computer Vision and Pattern Recognition
  • Artificial Intelligence

Cite this

Incremental commute time and its online applications. / Khoa, Nguyen Lu Dang; Wang, Yang; Chawla, Sanjay.

In: Pattern Recognition, Vol. 88, 01.04.2019, p. 101-112.

Research output: Contribution to journalArticle

Khoa, Nguyen Lu Dang ; Wang, Yang ; Chawla, Sanjay. / Incremental commute time and its online applications. In: Pattern Recognition. 2019 ; Vol. 88. pp. 101-112.
@article{92027fe396ed4cbc85caf7f786e8fb0d,
title = "Incremental commute time and its online applications",
abstract = "Commute time is a robust measure on graphs based on random walks. It has been successfully applied in many application domains including personalized search, collaborative filtering and network intrusion detection. However, the computation of the commute time is expensive since it involves the eigen decomposition of the graph Laplacian matrix. There has been effort to approximate the commute time but they only work in an offline mode. In this work, an accurate and efficient approximation for computing the commute time is proposed in an incremental fashion in order to facilitate online applications. Using the incremental commutime, we design an online anomaly detection application where the commute time of each new arriving data point to any point in the current graph can be estimated in constant time. The proposed approach shows its high accuracy and efficiency in synthetic and real datasets for online applications. It 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. We also discuss the use of incremental commute time for other online applications such as classification, graph ranking and clustering.",
keywords = "Anomaly detection, Commute time, Manifold learning, Online learning, Random walks",
author = "Khoa, {Nguyen Lu Dang} and Yang Wang and Sanjay Chawla",
year = "2019",
month = "4",
day = "1",
doi = "10.1016/j.patcog.2018.11.012",
language = "English",
volume = "88",
pages = "101--112",
journal = "Pattern Recognition",
issn = "0031-3203",
publisher = "Elsevier Limited",

}

TY - JOUR

T1 - Incremental commute time and its online applications

AU - Khoa, Nguyen Lu Dang

AU - Wang, Yang

AU - Chawla, Sanjay

PY - 2019/4/1

Y1 - 2019/4/1

N2 - Commute time is a robust measure on graphs based on random walks. It has been successfully applied in many application domains including personalized search, collaborative filtering and network intrusion detection. However, the computation of the commute time is expensive since it involves the eigen decomposition of the graph Laplacian matrix. There has been effort to approximate the commute time but they only work in an offline mode. In this work, an accurate and efficient approximation for computing the commute time is proposed in an incremental fashion in order to facilitate online applications. Using the incremental commutime, we design an online anomaly detection application where the commute time of each new arriving data point to any point in the current graph can be estimated in constant time. The proposed approach shows its high accuracy and efficiency in synthetic and real datasets for online applications. It 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. We also discuss the use of incremental commute time for other online applications such as classification, graph ranking and clustering.

AB - Commute time is a robust measure on graphs based on random walks. It has been successfully applied in many application domains including personalized search, collaborative filtering and network intrusion detection. However, the computation of the commute time is expensive since it involves the eigen decomposition of the graph Laplacian matrix. There has been effort to approximate the commute time but they only work in an offline mode. In this work, an accurate and efficient approximation for computing the commute time is proposed in an incremental fashion in order to facilitate online applications. Using the incremental commutime, we design an online anomaly detection application where the commute time of each new arriving data point to any point in the current graph can be estimated in constant time. The proposed approach shows its high accuracy and efficiency in synthetic and real datasets for online applications. It 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. We also discuss the use of incremental commute time for other online applications such as classification, graph ranking and clustering.

KW - Anomaly detection

KW - Commute time

KW - Manifold learning

KW - Online learning

KW - Random walks

UR - http://www.scopus.com/inward/record.url?scp=85056885073&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85056885073&partnerID=8YFLogxK

U2 - 10.1016/j.patcog.2018.11.012

DO - 10.1016/j.patcog.2018.11.012

M3 - Article

AN - SCOPUS:85056885073

VL - 88

SP - 101

EP - 112

JO - Pattern Recognition

JF - Pattern Recognition

SN - 0031-3203

ER -