Combination of gated recurrent unit and network in network for underwater acoustic target recognition

Shuang Yang, Xiangyang Zeng

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

2 Scopus citations

Abstract

Underwater acoustic target recognition is an important technical support for underwater acoustic information acquisition and confrontation. Taking the gated recurrent unit (GRU) into account which has an internal feedback mechanism reflecting the temporal correlation of underwater acoustic target features, a model combined with GRU and network in network (NIN) is proposed to recognize underwater acoustic targets. The proposed model integrates multi-channel information while compressing the hidden states of GRU, which enhances nonlinear fitting ability and local modeling ability of the network. Experiments based on measured underwater acoustic target signals show that the proposed model can achieve better classification performance than the multilayer stacked GRU model.

Original languageEnglish
Title of host publicationProceedings of INTER-NOISE 2021 - 2021 International Congress and Exposition of Noise Control Engineering
EditorsTyler Dare, Stuart Bolton, Patricia Davies, Yutong Xue, Gordon Ebbitt
PublisherThe Institute of Noise Control Engineering of the USA, Inc.
ISBN (Electronic)9781732598652
DOIs
StatePublished - 2021
Event50th International Congress and Exposition of Noise Control Engineering, INTER-NOISE 2021 - Washington, United States
Duration: 1 Aug 20215 Aug 2021

Publication series

NameProceedings of INTER-NOISE 2021 - 2021 International Congress and Exposition of Noise Control Engineering

Conference

Conference50th International Congress and Exposition of Noise Control Engineering, INTER-NOISE 2021
Country/TerritoryUnited States
CityWashington
Period1/08/215/08/21

Fingerprint

Dive into the research topics of 'Combination of gated recurrent unit and network in network for underwater acoustic target recognition'. Together they form a unique fingerprint.

Cite this