2021 6th International Conference on Machine Learning Technologies | 2021
Music Exploration Based on XGBoost Algorithm and Node2vec-BP Model
Abstract
The development trend of music genres tends to be homogenous, due to the reduction of music production costs, changes in market demand and the consolidation of composers’ thinking. In this article, the significance of each feature of music is inferred through the XGBoost algorithm, and further evidence of music homogeneity is obtained from the time series analysis of feature decomposition. In addition, we propose a high-performance music prediction system that combines Node2vec and BP algorithm to consider the influence of composers on the music revolution. It is concluded that the Node2vec-BP algorithm has higher generalization ability and accuracy than the traditional BP algorithm.