Inferring useful heuristics from the dynamics of iterative relational classifiers

Aram Galstyan, Paul R Cohen

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

5 Citations (Scopus)

Abstract

In this paper we consider dynamical properties of simple iterative relational classifiers. We conjecture that for a class of algorithms that use label-propagation the iterative procedure can lead to non-trivial dynamics in the number of newly classified instances. The underlaying reason for this non-triviality is that in relational networks true class labels are likely to propagate faster than false ones. We suggest that this phenomenon, which we call two-tiered dynamics for binary classifiers, can be used for establishing a self-consistent classification threshold and a criterion for stopping iteration. We demonstrate this effect for two unrelated binary classification problems using a variation of a iterative relational neighbor classifier. We also study analytically the dynamical properties of the suggested classifier, and compare its results to the numerical experiments on synthetic data.

Original languageEnglish (US)
Title of host publicationIJCAI International Joint Conference on Artificial Intelligence
Pages708-713
Number of pages6
StatePublished - 2005
Externally publishedYes
Event19th International Joint Conference on Artificial Intelligence, IJCAI 2005 - Edinburgh, United Kingdom
Duration: Jul 30 2005Aug 5 2005

Other

Other19th International Joint Conference on Artificial Intelligence, IJCAI 2005
CountryUnited Kingdom
CityEdinburgh
Period7/30/058/5/05

Fingerprint

Classifiers
Labels
Experiments

ASJC Scopus subject areas

  • Artificial Intelligence

Cite this

Galstyan, A., & Cohen, P. R. (2005). Inferring useful heuristics from the dynamics of iterative relational classifiers. In IJCAI International Joint Conference on Artificial Intelligence (pp. 708-713)

Inferring useful heuristics from the dynamics of iterative relational classifiers. / Galstyan, Aram; Cohen, Paul R.

IJCAI International Joint Conference on Artificial Intelligence. 2005. p. 708-713.

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

Galstyan, A & Cohen, PR 2005, Inferring useful heuristics from the dynamics of iterative relational classifiers. in IJCAI International Joint Conference on Artificial Intelligence. pp. 708-713, 19th International Joint Conference on Artificial Intelligence, IJCAI 2005, Edinburgh, United Kingdom, 7/30/05.
Galstyan A, Cohen PR. Inferring useful heuristics from the dynamics of iterative relational classifiers. In IJCAI International Joint Conference on Artificial Intelligence. 2005. p. 708-713
Galstyan, Aram ; Cohen, Paul R. / Inferring useful heuristics from the dynamics of iterative relational classifiers. IJCAI International Joint Conference on Artificial Intelligence. 2005. pp. 708-713
@inproceedings{4c1639e944794f218498fc5804fbed22,
title = "Inferring useful heuristics from the dynamics of iterative relational classifiers",
abstract = "In this paper we consider dynamical properties of simple iterative relational classifiers. We conjecture that for a class of algorithms that use label-propagation the iterative procedure can lead to non-trivial dynamics in the number of newly classified instances. The underlaying reason for this non-triviality is that in relational networks true class labels are likely to propagate faster than false ones. We suggest that this phenomenon, which we call two-tiered dynamics for binary classifiers, can be used for establishing a self-consistent classification threshold and a criterion for stopping iteration. We demonstrate this effect for two unrelated binary classification problems using a variation of a iterative relational neighbor classifier. We also study analytically the dynamical properties of the suggested classifier, and compare its results to the numerical experiments on synthetic data.",
author = "Aram Galstyan and Cohen, {Paul R}",
year = "2005",
language = "English (US)",
pages = "708--713",
booktitle = "IJCAI International Joint Conference on Artificial Intelligence",

}

TY - GEN

T1 - Inferring useful heuristics from the dynamics of iterative relational classifiers

AU - Galstyan, Aram

AU - Cohen, Paul R

PY - 2005

Y1 - 2005

N2 - In this paper we consider dynamical properties of simple iterative relational classifiers. We conjecture that for a class of algorithms that use label-propagation the iterative procedure can lead to non-trivial dynamics in the number of newly classified instances. The underlaying reason for this non-triviality is that in relational networks true class labels are likely to propagate faster than false ones. We suggest that this phenomenon, which we call two-tiered dynamics for binary classifiers, can be used for establishing a self-consistent classification threshold and a criterion for stopping iteration. We demonstrate this effect for two unrelated binary classification problems using a variation of a iterative relational neighbor classifier. We also study analytically the dynamical properties of the suggested classifier, and compare its results to the numerical experiments on synthetic data.

AB - In this paper we consider dynamical properties of simple iterative relational classifiers. We conjecture that for a class of algorithms that use label-propagation the iterative procedure can lead to non-trivial dynamics in the number of newly classified instances. The underlaying reason for this non-triviality is that in relational networks true class labels are likely to propagate faster than false ones. We suggest that this phenomenon, which we call two-tiered dynamics for binary classifiers, can be used for establishing a self-consistent classification threshold and a criterion for stopping iteration. We demonstrate this effect for two unrelated binary classification problems using a variation of a iterative relational neighbor classifier. We also study analytically the dynamical properties of the suggested classifier, and compare its results to the numerical experiments on synthetic data.

UR - http://www.scopus.com/inward/record.url?scp=84880759180&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84880759180&partnerID=8YFLogxK

M3 - Conference contribution

AN - SCOPUS:84880759180

SP - 708

EP - 713

BT - IJCAI International Joint Conference on Artificial Intelligence

ER -