Relational classification through three-state epidemie dynamics

Aram Galstyan, Paul R Cohen

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

3 Citations (Scopus)

Abstract

Relational classification in networked data plays an important role in many problems such as text categorization, classification of web pages, group finding in peer networks, etc. We have previously demonstrated that for a class of label propagating algorithms the underlying dynamics can be modeled as a two-state epidemic process on heterogeneous networks, where infected nodes correspond to classified data instances. We have also suggested a binary classification algorithm that utilizes non-trivial characteristics of epidemic dynamics. In this paper we extend our previous work by considering a three-state epidemic model for label propagation. Specifically, we introduce a new, intermediate state that corresponds to "susceptible" data instances. The utility of the added state is that it allows to control the rates of epidemic spreading, hence making the algorithm more flexible. We show empirically that this extension improves significantly the performance of the algorithm. In particular, we demonstrate that the new algorithm achieves good classification accuracy even for relatively large overlap across the classes.

Original languageEnglish (US)
Title of host publication2006 9th International Conference on Information Fusion, FUSION
DOIs
StatePublished - 2006
Externally publishedYes
Event2006 9th International Conference on Information Fusion, FUSION - Florence, Italy
Duration: Jul 10 2006Jul 13 2006

Other

Other2006 9th International Conference on Information Fusion, FUSION
CountryItaly
CityFlorence
Period7/10/067/13/06

Fingerprint

Labels
Heterogeneous networks
Websites

Keywords

  • Binary classification
  • Relational learning

ASJC Scopus subject areas

  • Computer Science Applications
  • Information Systems

Cite this

Galstyan, A., & Cohen, P. R. (2006). Relational classification through three-state epidemie dynamics. In 2006 9th International Conference on Information Fusion, FUSION [4085974] https://doi.org/10.1109/ICIF.2006.301688

Relational classification through three-state epidemie dynamics. / Galstyan, Aram; Cohen, Paul R.

2006 9th International Conference on Information Fusion, FUSION. 2006. 4085974.

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

Galstyan, A & Cohen, PR 2006, Relational classification through three-state epidemie dynamics. in 2006 9th International Conference on Information Fusion, FUSION., 4085974, 2006 9th International Conference on Information Fusion, FUSION, Florence, Italy, 7/10/06. https://doi.org/10.1109/ICIF.2006.301688
Galstyan A, Cohen PR. Relational classification through three-state epidemie dynamics. In 2006 9th International Conference on Information Fusion, FUSION. 2006. 4085974 https://doi.org/10.1109/ICIF.2006.301688
Galstyan, Aram ; Cohen, Paul R. / Relational classification through three-state epidemie dynamics. 2006 9th International Conference on Information Fusion, FUSION. 2006.
@inproceedings{8fe14d4fdd1147db8d1deb380cf773d0,
title = "Relational classification through three-state epidemie dynamics",
abstract = "Relational classification in networked data plays an important role in many problems such as text categorization, classification of web pages, group finding in peer networks, etc. We have previously demonstrated that for a class of label propagating algorithms the underlying dynamics can be modeled as a two-state epidemic process on heterogeneous networks, where infected nodes correspond to classified data instances. We have also suggested a binary classification algorithm that utilizes non-trivial characteristics of epidemic dynamics. In this paper we extend our previous work by considering a three-state epidemic model for label propagation. Specifically, we introduce a new, intermediate state that corresponds to {"}susceptible{"} data instances. The utility of the added state is that it allows to control the rates of epidemic spreading, hence making the algorithm more flexible. We show empirically that this extension improves significantly the performance of the algorithm. In particular, we demonstrate that the new algorithm achieves good classification accuracy even for relatively large overlap across the classes.",
keywords = "Binary classification, Relational learning",
author = "Aram Galstyan and Cohen, {Paul R}",
year = "2006",
doi = "10.1109/ICIF.2006.301688",
language = "English (US)",
isbn = "1424409535",
booktitle = "2006 9th International Conference on Information Fusion, FUSION",

}

TY - GEN

T1 - Relational classification through three-state epidemie dynamics

AU - Galstyan, Aram

AU - Cohen, Paul R

PY - 2006

Y1 - 2006

N2 - Relational classification in networked data plays an important role in many problems such as text categorization, classification of web pages, group finding in peer networks, etc. We have previously demonstrated that for a class of label propagating algorithms the underlying dynamics can be modeled as a two-state epidemic process on heterogeneous networks, where infected nodes correspond to classified data instances. We have also suggested a binary classification algorithm that utilizes non-trivial characteristics of epidemic dynamics. In this paper we extend our previous work by considering a three-state epidemic model for label propagation. Specifically, we introduce a new, intermediate state that corresponds to "susceptible" data instances. The utility of the added state is that it allows to control the rates of epidemic spreading, hence making the algorithm more flexible. We show empirically that this extension improves significantly the performance of the algorithm. In particular, we demonstrate that the new algorithm achieves good classification accuracy even for relatively large overlap across the classes.

AB - Relational classification in networked data plays an important role in many problems such as text categorization, classification of web pages, group finding in peer networks, etc. We have previously demonstrated that for a class of label propagating algorithms the underlying dynamics can be modeled as a two-state epidemic process on heterogeneous networks, where infected nodes correspond to classified data instances. We have also suggested a binary classification algorithm that utilizes non-trivial characteristics of epidemic dynamics. In this paper we extend our previous work by considering a three-state epidemic model for label propagation. Specifically, we introduce a new, intermediate state that corresponds to "susceptible" data instances. The utility of the added state is that it allows to control the rates of epidemic spreading, hence making the algorithm more flexible. We show empirically that this extension improves significantly the performance of the algorithm. In particular, we demonstrate that the new algorithm achieves good classification accuracy even for relatively large overlap across the classes.

KW - Binary classification

KW - Relational learning

UR - http://www.scopus.com/inward/record.url?scp=50149114115&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=50149114115&partnerID=8YFLogxK

U2 - 10.1109/ICIF.2006.301688

DO - 10.1109/ICIF.2006.301688

M3 - Conference contribution

AN - SCOPUS:50149114115

SN - 1424409535

SN - 9781424409532

BT - 2006 9th International Conference on Information Fusion, FUSION

ER -