publication . Article . 2013

Revisiting Inter-Genre Similarity

Bob Sturm;
Restricted English
  • Published: 01 Nov 2013
  • Country: Denmark
Abstract
We revisit the idea of ``inter-genre similarity'' (IGS)<br/>for machine learning in general,<br/>and music genre recognition in particular.<br/>We show analytically that the probability of error for IGS<br/>is higher than naive Bayes classification with zero-one loss (NB).<br/>We show empirically that IGS does not perform well,<br/>even for data that satisfies all its assumptions.
Subjects
free text keywords: Signal Processing, Electrical and Electronic Engineering, Applied Mathematics, Probability of error, Naive Bayes classifier, Computer science, Artificial intelligence, business.industry, business, Machine learning, computer.software_genre, computer, Bayes error rate, Pattern recognition
Related Organizations
Powered by OpenAIRE Research Graph
Any information missing or wrong?Report an Issue