Nonnegative Tucker Decomposition with Beta-divergence for Music Structure Analysis of Audio Signals

Valentin Leplat, Axel Marmoret, Florian Voorwinden, Frédéric Bimbot, Jérémy E. Cohen

Publication date: 27 October 2021

Abstract: Nonnegative Tucker decomposition (NTD), a tensor decomposition model, has received increased interest in the recent years because of its ability to blindly extract meaningful patterns, in particular in Music Information Retrieval. Nevertheless, existing algorithms to compute NTD are mostly designed for the Euclidean loss. This work proposes a multiplicative updates algorithm to compute NTD with the beta-divergence loss, often considered a better loss for audio processing. We notably show how to implement efficiently the multiplicative rules using tensor algebra. Finally, we show on a music structure analysis task that unsupervised NTD fitted with beta-divergence loss outperforms earlier results obtained with the Euclidean loss.

Has companion code repository: https://gitlab.inria.fr/amarmore/nonnegative-factorization

Mathematics Subject Classification ID

Software, source code, etc. for problems pertaining to linear algebra (15-04)

This page was built for publication: Nonnegative Tucker Decomposition with Beta-divergence for Music Structure Analysis of Audio Signals

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q6381446)