Beyond Statistical Relations: Integrating Knowledge Relations into Style Correlations for Multi-Label Music Style Classification
This work addresses the underfitting issue in multi-label music style classification for music websites, representing an incremental improvement by combining knowledge and statistical relations.
The paper tackles the problem of underfitting in multi-label music style classification by integrating external knowledge relations with statistical style correlations, using a graph convolutional network to learn deep correlations, and reports significant outperformance over state-of-the-art methods.
Automatically labeling multiple styles for every song is a comprehensive application in all kinds of music websites. Recently, some researches explore review-driven multi-label music style classification and exploit style correlations for this task. However, their methods focus on mining the statistical relations between different music styles and only consider shallow style relations. Moreover, these statistical relations suffer from the underfitting problem because some music styles have little training data. To tackle these problems, we propose a novel knowledge relations integrated framework (KRF) to capture the complete style correlations, which jointly exploits the inherent relations between music styles according to external knowledge and their statistical relations. Based on the two types of relations, we use a graph convolutional network to learn the deep correlations between styles automatically. Experimental results show that our framework significantly outperforms state-of-the-art methods. Further studies demonstrate that our framework can effectively alleviate the underfitting problem and learn meaningful style correlations. The source code can be available at https://github.com/Makwen1995/MusicGenre.