ITU

Analysis of the Effect of Dataset Construction Methodology on Transferability of Music Emotion Recognition Models

Research output: Conference Article in Proceeding or Book/Report chapterArticle in proceedingsResearchpeer-review

Standard

Analysis of the Effect of Dataset Construction Methodology on Transferability of Music Emotion Recognition Models. / Hult, Sabina; Kreiberg, Line Bay; Brandt, Sami Sebastian; Jónsson, Björn Thór.

Proceedings of the ACM International Conference on Multimedia Retrieval (ICMR). ed. / Cathal Gurrin; Björn Þór Jónsson; Noriko Kando; Klaus Schöffmann; Yi-Ping Phoebe Chen; Noel E. O'Connor. Dublin, Ireland : Association for Computing Machinery, 2020. p. 316-320.

Research output: Conference Article in Proceeding or Book/Report chapterArticle in proceedingsResearchpeer-review

Harvard

Hult, S, Kreiberg, LB, Brandt, SS & Jónsson, BT 2020, Analysis of the Effect of Dataset Construction Methodology on Transferability of Music Emotion Recognition Models. in C Gurrin, BÞ Jónsson, N Kando, K Schöffmann, Y-PP Chen & NE O'Connor (eds), Proceedings of the ACM International Conference on Multimedia Retrieval (ICMR). Association for Computing Machinery, Dublin, Ireland, pp. 316-320. https://doi.org/10.1145/3372278.3390733

APA

Hult, S., Kreiberg, L. B., Brandt, S. S., & Jónsson, B. T. (2020). Analysis of the Effect of Dataset Construction Methodology on Transferability of Music Emotion Recognition Models. In C. Gurrin, B. Þ. Jónsson, N. Kando, K. Schöffmann, Y-P. P. Chen, & N. E. O'Connor (Eds.), Proceedings of the ACM International Conference on Multimedia Retrieval (ICMR) (pp. 316-320). Association for Computing Machinery. https://doi.org/10.1145/3372278.3390733

Vancouver

Hult S, Kreiberg LB, Brandt SS, Jónsson BT. Analysis of the Effect of Dataset Construction Methodology on Transferability of Music Emotion Recognition Models. In Gurrin C, Jónsson BÞ, Kando N, Schöffmann K, Chen Y-PP, O'Connor NE, editors, Proceedings of the ACM International Conference on Multimedia Retrieval (ICMR). Dublin, Ireland: Association for Computing Machinery. 2020. p. 316-320 https://doi.org/10.1145/3372278.3390733

Author

Hult, Sabina ; Kreiberg, Line Bay ; Brandt, Sami Sebastian ; Jónsson, Björn Thór. / Analysis of the Effect of Dataset Construction Methodology on Transferability of Music Emotion Recognition Models. Proceedings of the ACM International Conference on Multimedia Retrieval (ICMR). editor / Cathal Gurrin ; Björn Þór Jónsson ; Noriko Kando ; Klaus Schöffmann ; Yi-Ping Phoebe Chen ; Noel E. O'Connor. Dublin, Ireland : Association for Computing Machinery, 2020. pp. 316-320

Bibtex

@inproceedings{8711cf95286245c88a3627666544d955,
title = "Analysis of the Effect of Dataset Construction Methodology on Transferability of Music Emotion Recognition Models",
abstract = "Indexing and retrieving music based on emotion is a powerful retrieval paradigm with many applications. Traditionally, studies in the field of music emotion recognition have focused on training and testing supervised machine learning models using a single music dataset. To be useful for today{\textquoteright}s vast music libraries, however, such machine learning models must be widely applicable beyond the dataset for which they were created. In this work, we analyze to what extent models trained on one music dataset can predict emotion in another dataset constructed using a different methodology, by conducting cross-dataset experiments with three publicly available datasets. Our results suggest that training a prediction model on a homogeneous dataset with carefully collected emotion annotations yields a better foundation than prediction models learned on a larger, more varied dataset, with less reliable annotations. ",
keywords = "Music emotion recognition, Cross-dataset, Model transferability",
author = "Sabina Hult and Kreiberg, {Line Bay} and Brandt, {Sami Sebastian} and J{\'o}nsson, {Bj{\"o}rn Th{\'o}r}",
year = "2020",
month = jun,
doi = "10.1145/3372278.3390733",
language = "English",
pages = "316--320",
editor = "Cathal Gurrin and J{\'o}nsson, {Bj{\"o}rn {\TH}{\'o}r} and Noriko Kando and Klaus Sch{\"o}ffmann and Chen, {Yi-Ping Phoebe} and O'Connor, {Noel E.}",
booktitle = "Proceedings of the ACM International Conference on Multimedia Retrieval (ICMR)",
publisher = "Association for Computing Machinery",
address = "United States",

}

RIS

TY - GEN

T1 - Analysis of the Effect of Dataset Construction Methodology on Transferability of Music Emotion Recognition Models

AU - Hult, Sabina

AU - Kreiberg, Line Bay

AU - Brandt, Sami Sebastian

AU - Jónsson, Björn Thór

PY - 2020/6

Y1 - 2020/6

N2 - Indexing and retrieving music based on emotion is a powerful retrieval paradigm with many applications. Traditionally, studies in the field of music emotion recognition have focused on training and testing supervised machine learning models using a single music dataset. To be useful for today’s vast music libraries, however, such machine learning models must be widely applicable beyond the dataset for which they were created. In this work, we analyze to what extent models trained on one music dataset can predict emotion in another dataset constructed using a different methodology, by conducting cross-dataset experiments with three publicly available datasets. Our results suggest that training a prediction model on a homogeneous dataset with carefully collected emotion annotations yields a better foundation than prediction models learned on a larger, more varied dataset, with less reliable annotations.

AB - Indexing and retrieving music based on emotion is a powerful retrieval paradigm with many applications. Traditionally, studies in the field of music emotion recognition have focused on training and testing supervised machine learning models using a single music dataset. To be useful for today’s vast music libraries, however, such machine learning models must be widely applicable beyond the dataset for which they were created. In this work, we analyze to what extent models trained on one music dataset can predict emotion in another dataset constructed using a different methodology, by conducting cross-dataset experiments with three publicly available datasets. Our results suggest that training a prediction model on a homogeneous dataset with carefully collected emotion annotations yields a better foundation than prediction models learned on a larger, more varied dataset, with less reliable annotations.

KW - Music emotion recognition

KW - Cross-dataset

KW - Model transferability

U2 - 10.1145/3372278.3390733

DO - 10.1145/3372278.3390733

M3 - Article in proceedings

SP - 316

EP - 320

BT - Proceedings of the ACM International Conference on Multimedia Retrieval (ICMR)

A2 - Gurrin, Cathal

A2 - Jónsson, Björn Þór

A2 - Kando, Noriko

A2 - Schöffmann, Klaus

A2 - Chen, Yi-Ping Phoebe

A2 - O'Connor, Noel E.

PB - Association for Computing Machinery

CY - Dublin, Ireland

ER -

ID: 85596862