PSO Based SVM as an Optimal Classifier for Classification of Radar Returns from Ionosphere
Jayshri D. Dhande · 2011
The aim of this paper is two fold. First, we present a through experimental study the different Artificial Neural Networks classifier for classification of radar returns from Ionosphere dataset. Second, we propose a novel classification system based on particle swarm optimization (PSO) to improve the generalization performance of the SVM classifier. For this purpose, we have optimized the kernel parameters of SVM classifier. The experiments were conducted on Jonhs Hopkins Ionosphere database. The comparison of different Neural Networks classifier and PSO-SVM is done based on Ionosphere dataset from UCI machine learning repository. The results show that RBFNN typically provide better classification results. When comparing to techniques applied to binary classification problems. Also SVM Classifier with RBF kernel gives best classification accuracy on training set. And PSO-SVM classifier with optimized kernel parameter selection for classification of radar returns from ionosphere dataset gives better accuracy and improves the generalization performance.