Hybrid GA-mSVM: Dimensionality reduction using hybrid genetic algorithm and modified support vector machine classifier

Abstract

<p>The expansion of technology results in the generation of enormous amounts of data in all areas. The researchers face a difficult challenge when they attempt to categorize these high-dimensional data. A method called as feature selection is used to reduce the high-dimensionality of the data. It is possible to consider selecting features as an issue of global combinatorial optimization in the field of machine learning. This minimizes on the total number of features, gets rid of data that is insignificant, noisy, or duplicative, and ultimately achieves an acceptable level of classification accuracy. For the purpose of dimensionality reduction, a unique approach known as the Hybrid Genetic Algorithm – modified Support Vector Machine Classifier (Hybrid GA-mSVM) is presented in this study. The genetic algorithm component conducts a search using the principles underlying the evolutionary process in order to find the best feature set and after that the trimmed dataset is provided to the SVMs. The results of the experiments reveal that the proposed approach efficiently minimizes the features and achieves a better classification accuracy than other feature selection methods.</p>