Universiti Teknologi Malaysia Institutional Repository

A significant feature selection in the Mahalanobis Taguchi system using modified-bees algorithm

Ramlie, Faizir and Muhamad, W. and Jamaludin, Khairur Rijal and Cudney, Elizabeth A. and Dollah, R. (2020) A significant feature selection in the Mahalanobis Taguchi system using modified-bees algorithm. International Journal of Engineering Research and Technology, 13 (1). pp. 117-136. ISSN 0974-3154

Full text not available from this repository.

Official URL: http://www.irphouse.com/ijert20/ijertv13n1_15.pdf


This paper compares the performance of orthogonal array (OA), modified-Bees Algorithm (mBA) and conventional Bees Algorithm (BA) in significant feature selection scheme (optimization) of the Mahalanobis-Taguchi System (MTS) methodology. The main contribution of this work is to address both performances in terms of computing cost i.e. computing time as well as classification accuracy rate. Several studies have been conducted to evaluate the performance of OA against other heuristic search techniques in MTS methodology however, discussions in terms of the computing speed performances were found to be lacking. Instead, the accuracy performances were given the emphasis by drawing criticisms towards the deployment of OA as ineffective as compared to other state-of-the-art heuristic algorithms. Bees Algorithm (BA) is one heuristic search technique that discovers optimal (or near optimal) solutions using search strategy mimics the social behaviour of a honeybee colony. In this comparison work, modified-BA (mBA) is introduced into the optimization scheme of MTS with a modification on its neighbourhood search mechanism from the original BA. Instead of searching in random mode, a backward selection method is proposed. MD is used as the result assessment metric while the larger-the-better type of SNR is deployed as the algorithm's objective function. The historical heart liver disease data are used as the case study on which the comparisons between OA, mBA and BA performances specifically in terms of the computing speed are made and addressed. The outcomes showed a promising performance of the mBA as compared to OA with a comparable classification accuracy rate. Eventhough OA outperformed mBA in terms of computational speed, the MTS manage to classify at the expense of lower number of variables suggested by mBA. The mBA also converges faster than the conventional BA in finding the potential solution of the case problem.

Item Type:Article
Uncontrolled Keywords:Bees Algorithm, Feature Selection
Subjects:T Technology > T Technology (General)
Divisions:Razak School of Engineering and Advanced Technology
ID Code:87257
Deposited By: Widya Wahid
Deposited On:31 Oct 2020 20:27
Last Modified:31 Oct 2020 20:27

Repository Staff Only: item control page