Comparative Analysis between Machine Learning Methods in Tones Classification
Ivelina Stefanova Balabanova, Valentina Ilieva Markova, Stela Savova Kostadinova, Georgi Georgiev · 2020
This paper presents an analysis between the indicators in synthesis of models based on machine learning techniques for RMS noise levels recognition of tones with different frequencies. Discriminant classification models were performed in MATLAB as pseudo-quadratic model with the highest accuracy of 84.650% was selected. Naïve Bayes algorithm with Gaussian and Kernel distributions is implemented in the classification process, as the better results were obtained in the second approach. In selection of the metric distance by the k-NN method an accuracy range from 89.800% to 91.050% is observed.