Efficient parameter selection of support vector machines

Abstract

Support Vector Machine (SVM) has, over the years established itself as an effective method for machine learning. SVM has strengths as such that it uses a kernel function to deal with arbitrary structured data which comprises of non-linear data sets. However, to fully optimize the benefits of using the kernel function, one will have to fine-tune the parameters of SVM in order to achieve feasible results. However, parameter selection can get complicated as the number of parameters and the size of the dataset increases. In this paper, we propose a method to deal with effective parameter selection for SVM for optimal performance through experiments done on heart sound data using the features of IEFE extraction technique. © 2008 Springer-Verlag.