An agent model for rough classifiers

Azuraliza Abu Bakar, Zulaiha Ali Othman, Abdul Razak Hamdan, R. Yusof, R. Ismail

Research output: Contribution to journalArticle

5 Citations (Scopus)

Abstract

This paper proposes a new agent-based approach in rough set classification theory. In data mining, the rough set technique is one classification technique. It generates rules from a large database and has mechanisms to handle noise and uncertainty in data. However, producing a rough classification model or rough classifier is computationally expensive, especially in its reduct computation phase: this is an NP-hard problem. These problems have brought about the generation of large amount of rules and high processing time. We solve these problems by embedding an agent-based algorithm within the rough modelling framework. In this study, the classifiers are based on creating agents within the main modelling processes such as reduct computation, rules generation and attribute projections. Four main agents are introduced: the interaction agent, weighted agent, reduction agent and default agent. We propose a heuristic for the default agent to control its searching activity. Experiments show that the proposed method significantly reduces the running time and the number of rules while maintaining the same classification accuracy.

Original languageEnglish
Pages (from-to)2239-2245
Number of pages7
JournalApplied Soft Computing Journal
Volume11
Issue number2
DOIs
Publication statusPublished - Mar 2011

Fingerprint

Classifiers
Data mining
Computational complexity
Processing
Experiments

Keywords

  • Agent models
  • Agent's interaction
  • Default rules
  • Reduction

ASJC Scopus subject areas

  • Software

Cite this

An agent model for rough classifiers. / Abu Bakar, Azuraliza; Ali Othman, Zulaiha; Hamdan, Abdul Razak; Yusof, R.; Ismail, R.

In: Applied Soft Computing Journal, Vol. 11, No. 2, 03.2011, p. 2239-2245.

Research output: Contribution to journalArticle

Abu Bakar, Azuraliza ; Ali Othman, Zulaiha ; Hamdan, Abdul Razak ; Yusof, R. ; Ismail, R. / An agent model for rough classifiers. In: Applied Soft Computing Journal. 2011 ; Vol. 11, No. 2. pp. 2239-2245.
@article{79cf0760ff28414396a5ea6f30278289,
title = "An agent model for rough classifiers",
abstract = "This paper proposes a new agent-based approach in rough set classification theory. In data mining, the rough set technique is one classification technique. It generates rules from a large database and has mechanisms to handle noise and uncertainty in data. However, producing a rough classification model or rough classifier is computationally expensive, especially in its reduct computation phase: this is an NP-hard problem. These problems have brought about the generation of large amount of rules and high processing time. We solve these problems by embedding an agent-based algorithm within the rough modelling framework. In this study, the classifiers are based on creating agents within the main modelling processes such as reduct computation, rules generation and attribute projections. Four main agents are introduced: the interaction agent, weighted agent, reduction agent and default agent. We propose a heuristic for the default agent to control its searching activity. Experiments show that the proposed method significantly reduces the running time and the number of rules while maintaining the same classification accuracy.",
keywords = "Agent models, Agent's interaction, Default rules, Reduction",
author = "{Abu Bakar}, Azuraliza and {Ali Othman}, Zulaiha and Hamdan, {Abdul Razak} and R. Yusof and R. Ismail",
year = "2011",
month = "3",
doi = "10.1016/j.asoc.2010.08.004",
language = "English",
volume = "11",
pages = "2239--2245",
journal = "Applied Soft Computing",
issn = "1568-4946",
publisher = "Elsevier BV",
number = "2",

}

TY - JOUR

T1 - An agent model for rough classifiers

AU - Abu Bakar, Azuraliza

AU - Ali Othman, Zulaiha

AU - Hamdan, Abdul Razak

AU - Yusof, R.

AU - Ismail, R.

PY - 2011/3

Y1 - 2011/3

N2 - This paper proposes a new agent-based approach in rough set classification theory. In data mining, the rough set technique is one classification technique. It generates rules from a large database and has mechanisms to handle noise and uncertainty in data. However, producing a rough classification model or rough classifier is computationally expensive, especially in its reduct computation phase: this is an NP-hard problem. These problems have brought about the generation of large amount of rules and high processing time. We solve these problems by embedding an agent-based algorithm within the rough modelling framework. In this study, the classifiers are based on creating agents within the main modelling processes such as reduct computation, rules generation and attribute projections. Four main agents are introduced: the interaction agent, weighted agent, reduction agent and default agent. We propose a heuristic for the default agent to control its searching activity. Experiments show that the proposed method significantly reduces the running time and the number of rules while maintaining the same classification accuracy.

AB - This paper proposes a new agent-based approach in rough set classification theory. In data mining, the rough set technique is one classification technique. It generates rules from a large database and has mechanisms to handle noise and uncertainty in data. However, producing a rough classification model or rough classifier is computationally expensive, especially in its reduct computation phase: this is an NP-hard problem. These problems have brought about the generation of large amount of rules and high processing time. We solve these problems by embedding an agent-based algorithm within the rough modelling framework. In this study, the classifiers are based on creating agents within the main modelling processes such as reduct computation, rules generation and attribute projections. Four main agents are introduced: the interaction agent, weighted agent, reduction agent and default agent. We propose a heuristic for the default agent to control its searching activity. Experiments show that the proposed method significantly reduces the running time and the number of rules while maintaining the same classification accuracy.

KW - Agent models

KW - Agent's interaction

KW - Default rules

KW - Reduction

UR - http://www.scopus.com/inward/record.url?scp=78751626004&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=78751626004&partnerID=8YFLogxK

U2 - 10.1016/j.asoc.2010.08.004

DO - 10.1016/j.asoc.2010.08.004

M3 - Article

AN - SCOPUS:78751626004

VL - 11

SP - 2239

EP - 2245

JO - Applied Soft Computing

JF - Applied Soft Computing

SN - 1568-4946

IS - 2

ER -