Study on Music Emotion Cognition Model Based on Applying the Improved Gene Expression Programming

Abstract

This paper proposes a music emotion cognition model by applying the improved genetic algorithm with dynamic mutation operator. Firstly, a Hevner emotion ring representing the music emotion space is presented by utilizing psychological fuzzy measure on words in music psychology. Secondly, the music emotion vector is introduced based on semantic similarity relation in computing with words. Then the mapping from high dimensional feature space of music to emotion space is built by using the genetic algorithm which can mine the emotion expressions. Finally the comparisons with some famous learning algorithms such as BP neural network, regression method with least square, show that the proposed method is effective for music emotion cognition.