{"title":"Wavelet Learning by Adaptive Hermite Cubic Splines applied to Bioacoustic Chirps","authors":"Randall Balestriero, H. Glotin","doi":"10.1109/OCEANSE.2019.8867410","DOIUrl":null,"url":null,"abstract":"Acoustic monitoring is used to study marine mammals in oceans. Automated analysis for captured sound is almost essential because of the large quantity of data. The deep learning approach is an efficient method, however acoustic features are often not adapted. Convolutional Neural Net can be seen as an optimal kernel decomposition, nevertheless it requires large amount of training data to learn its kernels. An alternative using pre-imposed kernels and thus not requiring any amount of data is the scattering framework which imposes as kernels wavelet filters. Our research focuses on adaptive time-frequency decomposition of bioacoustic signal, based on cubic spline learning representation. We give the theoretical derivations of the model, and demonstrates efficient real applications of various signal, including chirps of songs of Blue Whale.","PeriodicalId":375793,"journal":{"name":"OCEANS 2019 - Marseille","volume":"22 1","pages":"0"},"PeriodicalIF":0.0000,"publicationDate":"2019-06-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"2","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"OCEANS 2019 - Marseille","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.1109/OCEANSE.2019.8867410","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
引用次数: 2
Abstract
Acoustic monitoring is used to study marine mammals in oceans. Automated analysis for captured sound is almost essential because of the large quantity of data. The deep learning approach is an efficient method, however acoustic features are often not adapted. Convolutional Neural Net can be seen as an optimal kernel decomposition, nevertheless it requires large amount of training data to learn its kernels. An alternative using pre-imposed kernels and thus not requiring any amount of data is the scattering framework which imposes as kernels wavelet filters. Our research focuses on adaptive time-frequency decomposition of bioacoustic signal, based on cubic spline learning representation. We give the theoretical derivations of the model, and demonstrates efficient real applications of various signal, including chirps of songs of Blue Whale.