A machine learning pipeline for classification of cetacean echolocation clicks in large underwater acoustic datasets

Abstract

Machine learning algorithms, including recent advances in deep learning, are promising for tools for detection and classification of broadband high frequency signals in passive acoustic recordings. However, these methods are generally data-hungry and progress has been limited by challenges related to the lack of labeled datasets adequate for training and testing. Large quantities of known and as yet unidentified broadband signal types mingle in marine recordings, with variability introduced by acoustic propagation, source depths and orientations, and interacting signals. Manual classification of these datasets is unmanageable without an in-depth knowledge of the acoustic context of each recording location. A signal classification pipeline is presented which combines unsupervised and supervised learning phases with opportunities for expert oversight to label signals of interest. The method is illustrated with a case study using unsupervised clustering to identify five toothed whale echolocation click types and two anthropogenic signal categories. These categories are used to train a deep network to classify detected signals in either averaged time bins or as individual detections, in two independent datasets. Bin-level classification achieved higher overall precision (>99%) than click-level classification. However, click-level classification had the advantage of providing a label for every signal, and achieved higher overall recall, with overall precision from 92 to 94%. The results suggest that unsupervised learning is a viable solution for efficiently generating the large, representative training sets needed for applications of deep learning in passive acoustics.

The application of neural networks to classify dolphin echolocation clicks

“…Delphinid clicks are particularly challenging to classify due to their high variability, notably due to variable source levels and beam width, and rapid attenuation of the clicks (Au and Benoit-Bird, 2003; Finneran et al, 2016; Kloepper et al, 2012; Moore et al, 2008). Only nonsupervised networks have so far been tested successfully to classify delphinid clicks (Frasier, 2021; Frasier et al, 2017). Although this methodology was successful in separating group clicks into different classes, only one taxonomic group of delphinid was clearly identifiable (Risso’s dolphin) in the latest study (Frasier, 2021).…”

“…Only nonsupervised networks have so far been tested successfully to classify delphinid clicks (Frasier, 2021; Frasier et al, 2017). Although this methodology was successful in separating group clicks into different classes, only one taxonomic group of delphinid was clearly identifiable (Risso’s dolphin) in the latest study (Frasier, 2021). Here, we show that supervised networks allow click classification from taxonomically different species.…”

“…In parallel, RNNs are excellent for the extraction of patterns in time series or signals (LeCun et al, 2015) and have recently been used to classify sperm whale clicks to recognise vocal clans and individual whales (Bermant et al, 2019). Delphinid clicks have been previously classified successfully by unsupervised neural networks (Frasier, 2021;Frasier et al, 2016Frasier et al, , 2017, suggesting the ability of these networks to recognise different patterns and/or features in echolocation clicks. However, the classes could not all be identified as a taxonomic group (ie.…”

Machine Learning Approach for Predictive Maintenance of the Electrical Submersible Pumps (ESPs)

“…This requires the learning algorithm to generalize from the training data to unseen situations in a “reasonable” way. 40 …”

Caller ID for Risso’s and Pacific White-sided dolphins

“…The Lo pulsed signals were likely differentiated well because they were the most different from the other species in the dataset 20 and were not tested against Gg pulsed signals. Gg pulsed signals were included, though, in a deep network study recently 21 . Other strategies like combining wavelets and neural networks have differentiated between sperm whale and long-finned pilot whale clicks 22 .…”