Title: An efficient classification based on genetically optimised hybrid PCA-Kernel ELM learning

Authors: Tripti Goel; Vijay Nehra; Virendra P. Vishwakarma

Addresses: Department of Electronics and Communication Engineering, Bhagat Phool Singh Mahila Vishwavidyalaya, Khanpur Kalan, Sonepat, Haryana, India ' Department of Electronics and Communication Engineering, Bhagat Phool Singh Mahila Vishwavidyalaya, Khanpur Kalan, Sonepat, Haryana, India ' University School of Information and Communication Technology, Guru Gobind Singh Indraprastha University, Dwarka, Delhi, India

Abstract: A new classifier based on genetically optimised kernel extreme learning machine (KELM) is presented here. Firstly, principal component analysis (PCA) is used to retrieve the important features from the datasets and further these features are used for classification. Classification is done by using the genetically optimised KELM algorithm in which the kernel parameters of the kernel function of ELM are optimised by using the genetic algorithm (GA). The present approach is investigated on eight benchmark biomedical datasets from UCI machine learning repository and AT&T face database to show its efficiency and effectiveness. The results are validated by using classification accuracy, ROC and cross validation. The results show that the proposed learning algorithm is better in terms of generalisation performance and learning speed compared to other state of the art learning algorithms.

Keywords: artificial neural networks; ANNs; biomedical datasets; extreme learning machine; genetic algorithms; kernel ELM; principal component analysis; PCA; feature extraction; classification; machine learning; biomedical engineering.

DOI: 10.1504/IJAPR.2016.079735

International Journal of Applied Pattern Recognition, 2016 Vol.3 No.3, pp.241 - 258

Received: 07 Dec 2015
Accepted: 21 Mar 2016
Published online: 13 Oct 2016 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article