Universiti Teknologi Malaysia Institutional Repository

Development of a hybrid artificial neural network - naive bayes classifier for binary classification problem of imbalanced datasets

Adam, A. and Shapiai, Mohd. Ibrahim and Ibrahim, Zuwairie and Khalid, Marzuki and Jau, L. W. (2011) Development of a hybrid artificial neural network - naive bayes classifier for binary classification problem of imbalanced datasets. ICIC Express Letters, 5 (9). pp. 3171-3175. ISSN 1881-803X

[img] PDF (Abstract)
68Kb

Official URL: http://www.ijicic.org/el-5(9)a.htm

Abstract

This paper presents a hybrid approach that consists of two different methods from machine learning technique, which are the Artificial Neural Network (ANN) and Naïve Bayes. The proposed technique is purposely developed for classifying the two classes of imbalanced datasets. Architecture of ANN is based on a single layer feedforward ANN for binary classification and the learning algorithm is assisted by the Particle Swarm Optimization (PSO) algorithm. As a main classifier, the Naïve Bayes is still being kept by using a conventional method. Consequently, by comparing with the individual classifiers that are used in this paper, the proposed approach is capable of improving the prediction performance that is evaluated by geometric mean (Gmean) as the performance measure.

Item Type:Article
Uncontrolled Keywords:artificial neural network, binary classification, hybrid classifier, imbalanced dataset, naïve bayes, particle swarm optimization
Subjects:T Technology > TK Electrical engineering. Electronics Nuclear engineering
Divisions:Electrical Engineering
ID Code:28826
Deposited By: Liza Porijo
Deposited On:29 Nov 2012 06:21
Last Modified:29 Nov 2012 06:21

Repository Staff Only: item control page