Login (DCU Staff Only)
Login (DCU Staff Only)

DORAS | DCU Research Repository

Explore open access research and scholarly works from DCU

Advanced Search

Unsupervised contrastive learning of sound event representations

Fonseca, Eduardo orcid logoORCID: 0000-0001-9872-3917, Ortego, Diego orcid logoORCID: 0000-0002-1011-3610, McGuinness, Kevin orcid logoORCID: 0000-0003-1336-6477, O'Connor, Noel E. orcid logoORCID: 0000-0002-4033-9135 and Serra, Xavier orcid logoORCID: 0000-0003-1395-2345 (2021) Unsupervised contrastive learning of sound event representations. In: IEEE International Conference on Acoustics, Speech and Signal Processing, 6-11 June 2021, Toronto, Canada (Online).

Abstract
Self-supervised representation learning can mitigate the limitations in recognition tasks with few manually labeled data but abundant unlabeled data—a common scenario in sound event research. In this work, we explore unsupervised contrastive learning as a way to learn sound event representations. To this end, we propose to use the pretext task of contrasting differently augmented views of sound events. The views are computed primarily via mixing of training examples with unrelated backgrounds, followed by other data augmentations. We analyze the main components of our method via ablation experiments. We evaluate the learned representations using linear evaluation, and in two in-domain downstream sound event classification tasks, namely, using limited manually labeled data, and using noisy labeled data. Our results suggest that unsupervised contrastive pre-training can mitigate the impact of data scarcity and increase robustness against noisy labels.
Metadata
Item Type:Conference or Workshop Item (Paper)
Event Type:Conference
Refereed:Yes
Uncontrolled Keywords:Contrastive learning; sound event classification; audio representation learning; self-supervision
Subjects:Computer Science > Artificial intelligence
Computer Science > Image processing
Computer Science > Machine learning
Engineering > Acoustical engineering
DCU Faculties and Centres:DCU Faculties and Schools > Faculty of Engineering and Computing > School of Electronic Engineering
Research Initiatives and Centres > INSIGHT Centre for Data Analytics
Published in: 2021 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP). .
Official URL:https://doi.org/10.1109/ICASSP39728.2021.9415009
Copyright Information:© 2021 The Authors
Funders:Science Foundation Ireland (SFI/15/SIRG/3283), Young European Research University Network (YERUN), Google Faculty Research Award 2018
ID Code:25575
Deposited On:02 Mar 2021 11:56 by Diego Ortego Hernández . Last Modified 04 Nov 2021 13:36
Documents

Full text available as:

[thumbnail of 2021_ICASSP Unsupervised Contrastive Learning of Sound Event Representations.pdf]
Preview
PDF - Requires a PDF viewer such as GSview, Xpdf or Adobe Acrobat Reader
692kB
Downloads

Downloads

Downloads per month over past year

Archive Staff Only: edit this record