Semi-supervised learning helps in sound event classification

Zixing Zhang; Bjorn Schuller

doi:10.1109/ICASSP.2012.6287884

Semi-supervised learning helps in sound event classification

Zixing Zhang, Bjorn Schuller

Chair of Health Informatics

Technical University of Munich

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

42 Scopus citations

Abstract

We investigate the suitability of semi-supervised learning in sound event classification on a large database of 17 k sound clips. Seven categories are chosen based on the findsounds.com schema: animals, people, nature, vehicles, noisemakers, office, and musical instruments. Our results show that adding unlabelled sound event data to the training set based on sufficient classifier confidence level after its automatic labelling level can significantly enhance classification performance. Furthermore, combined with optimal re-sampling of originally labelled instances and iteratively learning in semi-supervised manner, the expected gain can reach approximately half the one achieved by using the originally manually labelled data. Overall, maximum performance of 71.7% can be reported for the automatic classification of sound in a large-scale archive.

Original language	English
Title of host publication	2012 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2012 - Proceedings
Pages	333-336
Number of pages	4
DOIs	https://doi.org/10.1109/ICASSP.2012.6287884
State	Published - 2012
Event	2012 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2012 - Kyoto, Japan Duration: 25 Mar 2012 → 30 Mar 2012

Publication series

Name	ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings
ISSN (Print)	1520-6149

Conference

Conference	2012 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2012
Country/Territory	Japan
City	Kyoto
Period	25/03/12 → 30/03/12

Keywords

Semi-supervised Learning
Sound Event Classification

Access to Document

10.1109/ICASSP.2012.6287884

Cite this

Zhang, Z., & Schuller, B. (2012). Semi-supervised learning helps in sound event classification. In 2012 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2012 - Proceedings (pp. 333-336). Article 6287884 (ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings). https://doi.org/10.1109/ICASSP.2012.6287884

@inproceedings{f64ac0850ff64d3691bb6b3f6aff2ccf,

title = "Semi-supervised learning helps in sound event classification",

abstract = "We investigate the suitability of semi-supervised learning in sound event classification on a large database of 17 k sound clips. Seven categories are chosen based on the findsounds.com schema: animals, people, nature, vehicles, noisemakers, office, and musical instruments. Our results show that adding unlabelled sound event data to the training set based on sufficient classifier confidence level after its automatic labelling level can significantly enhance classification performance. Furthermore, combined with optimal re-sampling of originally labelled instances and iteratively learning in semi-supervised manner, the expected gain can reach approximately half the one achieved by using the originally manually labelled data. Overall, maximum performance of 71.7% can be reported for the automatic classification of sound in a large-scale archive.",

keywords = "Semi-supervised Learning, Sound Event Classification",

author = "Zixing Zhang and Bjorn Schuller",

year = "2012",

doi = "10.1109/ICASSP.2012.6287884",

language = "English",

isbn = "9781467300469",

series = "ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings",

pages = "333--336",

booktitle = "2012 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2012 - Proceedings",

note = "2012 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2012 ; Conference date: 25-03-2012 Through 30-03-2012",

}

Zhang, Z & Schuller, B 2012, Semi-supervised learning helps in sound event classification. in 2012 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2012 - Proceedings., 6287884, ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings, pp. 333-336, 2012 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2012, Kyoto, Japan, 25/03/12. https://doi.org/10.1109/ICASSP.2012.6287884

Semi-supervised learning helps in sound event classification. / Zhang, Zixing; Schuller, Bjorn.
2012 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2012 - Proceedings. 2012. p. 333-336 6287884 (ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

TY - GEN

T1 - Semi-supervised learning helps in sound event classification

AU - Zhang, Zixing

AU - Schuller, Bjorn

PY - 2012

Y1 - 2012

N2 - We investigate the suitability of semi-supervised learning in sound event classification on a large database of 17 k sound clips. Seven categories are chosen based on the findsounds.com schema: animals, people, nature, vehicles, noisemakers, office, and musical instruments. Our results show that adding unlabelled sound event data to the training set based on sufficient classifier confidence level after its automatic labelling level can significantly enhance classification performance. Furthermore, combined with optimal re-sampling of originally labelled instances and iteratively learning in semi-supervised manner, the expected gain can reach approximately half the one achieved by using the originally manually labelled data. Overall, maximum performance of 71.7% can be reported for the automatic classification of sound in a large-scale archive.

AB - We investigate the suitability of semi-supervised learning in sound event classification on a large database of 17 k sound clips. Seven categories are chosen based on the findsounds.com schema: animals, people, nature, vehicles, noisemakers, office, and musical instruments. Our results show that adding unlabelled sound event data to the training set based on sufficient classifier confidence level after its automatic labelling level can significantly enhance classification performance. Furthermore, combined with optimal re-sampling of originally labelled instances and iteratively learning in semi-supervised manner, the expected gain can reach approximately half the one achieved by using the originally manually labelled data. Overall, maximum performance of 71.7% can be reported for the automatic classification of sound in a large-scale archive.

KW - Semi-supervised Learning

KW - Sound Event Classification

UR - http://www.scopus.com/inward/record.url?scp=84867599761&partnerID=8YFLogxK

U2 - 10.1109/ICASSP.2012.6287884

DO - 10.1109/ICASSP.2012.6287884

M3 - Conference contribution

AN - SCOPUS:84867599761

SN - 9781467300469

T3 - ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings

SP - 333

EP - 336

BT - 2012 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2012 - Proceedings

T2 - 2012 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2012

Y2 - 25 March 2012 through 30 March 2012

ER -

Semi-supervised learning helps in sound event classification

Abstract

Publication series

Conference

Keywords

Access to Document

Other files and links

Fingerprint

Cite this