A Double Weighted KNN Algorithm and Its Application in the Music Genre Classification

Abstract

This paper proposes a double weighted KNN algorithm, and applies it to the research of music genre automatic classification. This algorithm makes improvements in two aspects in the traditional KNN algorithm, which can effectively solve the problem that traditional KNN algorithm ignores the degree of correlation between attributes and categories in the classification process, and the problem that it only considers the number of the nearest samples and ignores the existence of similarity differences between the nearest samples and the samples to be classified in the process of category judgment, thus can effectively improve the classification accuracy. In this paper, this algorithm is applied to the music genre classification, and experiments prove that the algorithm can achieve higher classification accuracy in terms of music genre classification, and even has a better classification performance especially where there is no obvious difference between some of the categories and in the situation of crosscategory, and this algorithm is simple and symmetrical, with no complex dependency, high calculation efficiency, and adaptable to the demand of mass music data classification.