Konferenzbeitrag
Measuring the performance of evolutionary multi-objective feature selection for prediction of musical genres and styles
Lade...
Volltext URI
Dokumententyp
Text/Conference Paper
Dateien
Zusatzinformation
Datum
2013
Autor:innen
Zeitschriftentitel
ISSN der Zeitschrift
Bandtitel
Verlag
Gesellschaft für Informatik e.V.
Zusammenfassung
The prediction of high-level music categories, such as genres, styles, or personal preferences, helps to organise music collections. The relevance of single audio features for automatic classification depends on a certain category. Relevant feature subsets for each classification task can be identified by means of feature selec- tion. Continuing our previous studies on multi-objective feature selection for music classification, in this work we measure an impact of evolutionary multi-objective fea- ture selection on classification performance and compare it to the baseline application without feature selection. As confirmed by statistical tests, the integration of evolu- tionary multi-objective feature selection leads to a significant increase of performance according to both evaluation criteria as well as to classification error. This holds for all four tested classification methods and six music categories.