dc.contributor.author | Zhu, Xiaoyu | |
dc.contributor.author | Dong, Hefeng | |
dc.contributor.author | Salvo Rossi, Pierluigi | |
dc.contributor.author | Landrø, Martin | |
dc.date.accessioned | 2022-10-26T08:19:54Z | |
dc.date.available | 2022-10-26T08:19:54Z | |
dc.date.created | 2021-12-22T11:18:02Z | |
dc.date.issued | 2021 | |
dc.identifier.citation | Proceedings of IEEE Sensors. 2021, . | en_US |
dc.identifier.issn | 1930-0395 | |
dc.identifier.uri | https://hdl.handle.net/11250/3028336 | |
dc.description.abstract | This work introduces a two-step self-supervised learning scheme, namely contrastive predictive coding (CPC), for underwater source localization. In the first step, a CPC-based self-supervised feature extractor is trained with the acoustic signals. In the second step, the encoder with frozen parameters is taken from the trained feature extractor and connected with a multi-layer perceptron (MLP) trained for source localization on a small labeled dataset. This approach is evaluated on a public dataset, SWellEx-96 Event S5, against an autoencoder (AE) scheme and a purely supervised scheme. The results indicate that the CPC scheme has the best performance and can extract the slow-changing features related to the source. | en_US |
dc.language.iso | eng | en_US |
dc.publisher | Institute of Electrical and Electronics Engineers (IEEE) | en_US |
dc.title | Self-supervised Underwater Source Localization based on Contrastive Predictive Coding | en_US |
dc.type | Peer reviewed | en_US |
dc.type | Journal article | en_US |
dc.description.version | publishedVersion | en_US |
dc.rights.holder | This version of the article will not be available due to copyright restrictions by IEEE | en_US |
dc.source.pagenumber | 4 | en_US |
dc.source.journal | Proceedings of IEEE Sensors | en_US |
dc.identifier.doi | 10.1109/SENSORS47087.2021.9639566 | |
dc.identifier.cristin | 1971394 | |
dc.relation.project | Norges forskningsråd: 294404 | en_US |
dc.relation.project | Norges forskningsråd: 309960 | en_US |
cristin.ispublished | true | |
cristin.fulltext | original | |
cristin.qualitycode | 1 | |