Anzeige
Erschienen in:
2014 | OriginalPaper | Buchkapitel
From Personalized to Hierarchically Structured Classifiers for Retrieving Music by Mood
verfasst von : Amanda Cohen Mostafavi, Zbigniew W. Raś, Alicja A. Wieczorkowska
Erschienen in: New Frontiers in Mining Complex Patterns
Aktivieren Sie unsere intelligente Suche, um passende Fachinhalte oder Patente zu finden.
Wählen Sie Textabschnitte aus um mit Künstlicher Intelligenz passenden Patente zu finden. powered by
Markieren Sie Textabschnitte, um KI-gestützt weitere passende Inhalte zu finden. powered by
Abstract
With the increased amount of music that is available to the average user, either online or through their own collection, there is a need to develop new ways to organize and retrieve music. We propose a system by which we develop a set of personalized emotion classifiers, one for each emotion in a set of 16 and a set unique to each user. We train a set of emotion classifiers using feature data extracted from audio which has been tagged with a set of emotions by volunteers. We then develop SVM, kNN, Random Forest, and C4.5 tree based classifiers for each emotion and determine the best classification algorithm. We then compare our personalized emotion classifiers to a set of non-personalized classifiers. Finally, we present a method for efficiently developing personalized classifiers based on hierarchical clustering.