Abstract

Network modeling and analysis has become a fundamental tool for studying various complex systems. This paper proposes an extension of statistical monitoring to network streams, which is crucial for effective decision-making in various applications. To this end, a model for the probability of edge existence as a function of vertex attributes is constructed and a likelihood method is developed to monitor the underlying network model. The method is flexible to detect any form of anomaly that arises from different network edge-formation mechanisms. Experiments on simulated and real network streams depict the properties and benefits of the method compared with existing methods in the literature.

Original languageEnglish (US)
Pages (from-to)28-43
Number of pages16
JournalJournal of Quality Technology
Volume48
Issue number1
StatePublished - Jan 1 2016

Fingerprint

Monitoring
Large scale systems
Decision making
Experiments
Homogeneity
Network model
Complex systems
Anomaly
Experiment
Modeling

Keywords

  • Change detection
  • Dynamic network
  • Likelihood-ratio test

ASJC Scopus subject areas

  • Industrial and Manufacturing Engineering
  • Safety, Risk, Reliability and Quality
  • Strategy and Management
  • Management Science and Operations Research

Cite this

Monitoring temporal homogeneity in attributed network streams. / Azarnoush, Bahareh; Paynabar, Kamran; Bekki, Jennifer; Runger, George.

In: Journal of Quality Technology, Vol. 48, No. 1, 01.01.2016, p. 28-43.

Research output: Contribution to journalArticle

@article{fa1aa718003c4d71b85e8b82c6a6662b,
title = "Monitoring temporal homogeneity in attributed network streams",
abstract = "Network modeling and analysis has become a fundamental tool for studying various complex systems. This paper proposes an extension of statistical monitoring to network streams, which is crucial for effective decision-making in various applications. To this end, a model for the probability of edge existence as a function of vertex attributes is constructed and a likelihood method is developed to monitor the underlying network model. The method is flexible to detect any form of anomaly that arises from different network edge-formation mechanisms. Experiments on simulated and real network streams depict the properties and benefits of the method compared with existing methods in the literature.",
keywords = "Change detection, Dynamic network, Likelihood-ratio test",
author = "Bahareh Azarnoush and Kamran Paynabar and Jennifer Bekki and George Runger",
year = "2016",
month = "1",
day = "1",
language = "English (US)",
volume = "48",
pages = "28--43",
journal = "Journal of Quality Technology",
issn = "0022-4065",
publisher = "American Society for Quality",
number = "1",

}

TY - JOUR

T1 - Monitoring temporal homogeneity in attributed network streams

AU - Azarnoush, Bahareh

AU - Paynabar, Kamran

AU - Bekki, Jennifer

AU - Runger, George

PY - 2016/1/1

Y1 - 2016/1/1

N2 - Network modeling and analysis has become a fundamental tool for studying various complex systems. This paper proposes an extension of statistical monitoring to network streams, which is crucial for effective decision-making in various applications. To this end, a model for the probability of edge existence as a function of vertex attributes is constructed and a likelihood method is developed to monitor the underlying network model. The method is flexible to detect any form of anomaly that arises from different network edge-formation mechanisms. Experiments on simulated and real network streams depict the properties and benefits of the method compared with existing methods in the literature.

AB - Network modeling and analysis has become a fundamental tool for studying various complex systems. This paper proposes an extension of statistical monitoring to network streams, which is crucial for effective decision-making in various applications. To this end, a model for the probability of edge existence as a function of vertex attributes is constructed and a likelihood method is developed to monitor the underlying network model. The method is flexible to detect any form of anomaly that arises from different network edge-formation mechanisms. Experiments on simulated and real network streams depict the properties and benefits of the method compared with existing methods in the literature.

KW - Change detection

KW - Dynamic network

KW - Likelihood-ratio test

UR - http://www.scopus.com/inward/record.url?scp=84961140236&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84961140236&partnerID=8YFLogxK

M3 - Article

VL - 48

SP - 28

EP - 43

JO - Journal of Quality Technology

JF - Journal of Quality Technology

SN - 0022-4065

IS - 1

ER -