Fuzzy gaussian classifier for combining multiple learners

Farid Ali; Neamat El Gayar; Sanaa El Ola

Fuzzy gaussian classifier for combining multiple learners

Farid Ali^*, Neamat El Gayar, Sanaa El Ola

^*Corresponding author for this work

School of Mathematical & Computer Sciences

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

Abstract

In the field of pattern recognition multiple classifier systems based on the combination of outputs from different classifiers have been proposed as a method of high performance classification systems. The objective of this work is to develop a fuzzy Gaussian classifier for combining multiple learners, we use a fuzzy Gaussian model to combine the outputs obtained from K-nearest neighbor classifier (KNN), Fuzzy K-nearest neighbor classifier and Multi-layer Perceptron (MLP) and then compare the results with Fuzzy Integral, Decision Templates, Weighted Majority, Majority Naïve Bayes, Maximum, Minimum, Average and Product combination methods. Results on two benchmark data sets show that the proposed fusion method outperforms a wide variety of existing classifier combination methods.

Original language	English
Title of host publication	2010 The 7th International Conference on Informatics and Systems (INFOS)
Publisher	IEEE
ISBN (Electronic)	9781424458288
ISBN (Print)	9781424458288
Publication status	Published - 6 May 2010
Event	7th International Conference on Informatics and Systems 2010 - Cairo, Egypt Duration: 28 Mar 2010 → 30 Mar 2010

Conference

Conference	7th International Conference on Informatics and Systems 2010
Abbreviated title	INFOS 2010
Country/Territory	Egypt
City	Cairo
Period	28/03/10 → 30/03/10

Keywords

Classifier combination
Fuzzy gaussian classifier
Fuzzy K-nearest neighbors
K-nearest neighbors
Multi-layer perceptron

ASJC Scopus subject areas

Information Systems

Cite this

@inproceedings{fc9cf235b557436a8baa605f888356a5,

title = "Fuzzy gaussian classifier for combining multiple learners",

abstract = "In the field of pattern recognition multiple classifier systems based on the combination of outputs from different classifiers have been proposed as a method of high performance classification systems. The objective of this work is to develop a fuzzy Gaussian classifier for combining multiple learners, we use a fuzzy Gaussian model to combine the outputs obtained from K-nearest neighbor classifier (KNN), Fuzzy K-nearest neighbor classifier and Multi-layer Perceptron (MLP) and then compare the results with Fuzzy Integral, Decision Templates, Weighted Majority, Majority Na{\"i}ve Bayes, Maximum, Minimum, Average and Product combination methods. Results on two benchmark data sets show that the proposed fusion method outperforms a wide variety of existing classifier combination methods.",

keywords = "Classifier combination, Fuzzy gaussian classifier, Fuzzy K-nearest neighbors, K-nearest neighbors, Multi-layer perceptron",

author = "Farid Ali and {El Gayar}, Neamat and {El Ola}, Sanaa",

year = "2010",

month = may,

day = "6",

language = "English",

isbn = "9781424458288",

booktitle = "2010 The 7th International Conference on Informatics and Systems (INFOS)",

publisher = "IEEE",

address = "United States",

note = "7th International Conference on Informatics and Systems 2010, INFOS 2010 ; Conference date: 28-03-2010 Through 30-03-2010",

}

TY - GEN

T1 - Fuzzy gaussian classifier for combining multiple learners

AU - Ali, Farid

AU - El Gayar, Neamat

AU - El Ola, Sanaa

PY - 2010/5/6

Y1 - 2010/5/6

N2 - In the field of pattern recognition multiple classifier systems based on the combination of outputs from different classifiers have been proposed as a method of high performance classification systems. The objective of this work is to develop a fuzzy Gaussian classifier for combining multiple learners, we use a fuzzy Gaussian model to combine the outputs obtained from K-nearest neighbor classifier (KNN), Fuzzy K-nearest neighbor classifier and Multi-layer Perceptron (MLP) and then compare the results with Fuzzy Integral, Decision Templates, Weighted Majority, Majority Naïve Bayes, Maximum, Minimum, Average and Product combination methods. Results on two benchmark data sets show that the proposed fusion method outperforms a wide variety of existing classifier combination methods.

AB - In the field of pattern recognition multiple classifier systems based on the combination of outputs from different classifiers have been proposed as a method of high performance classification systems. The objective of this work is to develop a fuzzy Gaussian classifier for combining multiple learners, we use a fuzzy Gaussian model to combine the outputs obtained from K-nearest neighbor classifier (KNN), Fuzzy K-nearest neighbor classifier and Multi-layer Perceptron (MLP) and then compare the results with Fuzzy Integral, Decision Templates, Weighted Majority, Majority Naïve Bayes, Maximum, Minimum, Average and Product combination methods. Results on two benchmark data sets show that the proposed fusion method outperforms a wide variety of existing classifier combination methods.

KW - Classifier combination

KW - Fuzzy gaussian classifier

KW - Fuzzy K-nearest neighbors

KW - K-nearest neighbors

KW - Multi-layer perceptron

UR - http://www.scopus.com/inward/record.url?scp=77953204588&partnerID=8YFLogxK

M3 - Conference contribution

AN - SCOPUS:77953204588

SN - 9781424458288

BT - 2010 The 7th International Conference on Informatics and Systems (INFOS)

PB - IEEE

T2 - 7th International Conference on Informatics and Systems 2010

Y2 - 28 March 2010 through 30 March 2010

ER -

Fuzzy gaussian classifier for combining multiple learners

Abstract

Conference

Keywords

ASJC Scopus subject areas

Other files and links

Fingerprint

Cite this