The Application of Multiple Classifier System for Environmental Audio Classification

Abstract

Multiple classifier system trains different classifiers and combines their predictions to improve the accuracy of classification. This paper explains the popular algorithms and strategies in multiple classifier system, and points out the key factors to affect the performance of the application of multiple classifier system. The experiments are carried out on given environmental audio data in order to compare the singular classifier methods with multiple classifier system such as Random Forest and MCS, as well as Bagging and AdaBoost. The experimental results show that the multiple classifiers technology outperforms the singular classifier and obtains better performance in environmental audio data classification. It provides an effective way to guarantee the performance and generalization of classification.