Detecting and reducing heterogeneity of error in acoustic classification

Abstract Passive acoustic monitoring can be an effective method for monitoring species, allowing the assembly of large audio datasets, removing logistical constraints in data collection and reducing anthropogenic monitoring disturbances. However, the analysis of large acoustic datasets is challengin...

Full description

Saved in:
Bibliographic Details
Main Authors: Oliver C. Metcalf, Jos Barlow, Yves Bas, Erika Berenguer, Christian Devenish, Filipe França, Stuart Marsden, Charlotte Smith, Alexander C. Lees
Format: Article
Language:English
Published: Wiley 2022-11-01
Series:Methods in Ecology and Evolution
Subjects:
Online Access:https://doi.org/10.1111/2041-210X.13967
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1849771914942218240
author Oliver C. Metcalf
Jos Barlow
Yves Bas
Erika Berenguer
Christian Devenish
Filipe França
Stuart Marsden
Charlotte Smith
Alexander C. Lees
author_facet Oliver C. Metcalf
Jos Barlow
Yves Bas
Erika Berenguer
Christian Devenish
Filipe França
Stuart Marsden
Charlotte Smith
Alexander C. Lees
author_sort Oliver C. Metcalf
collection DOAJ
description Abstract Passive acoustic monitoring can be an effective method for monitoring species, allowing the assembly of large audio datasets, removing logistical constraints in data collection and reducing anthropogenic monitoring disturbances. However, the analysis of large acoustic datasets is challenging and fully automated machine learning processes are rarely developed or implemented in ecological field studies. One of the greatest uncertainties hindering the development of these methods is spatial generalisability—can an algorithm trained on data from one place be used elsewhere? We demonstrate that heterogeneity of error across space is a problem that could go undetected using common classification accuracy metrics. Second, we develop a method to assess the extent of heterogeneity of error in a random forest classification model for six Amazonian bird species. Finally, we propose two complementary ways to reduce heterogeneity of error, by (i) accounting for it in the thresholding process and (ii) using a secondary classifier that uses contextual data. We found that using a thresholding approach that accounted for heterogeneity of precision error reduced the coefficient of variation of the precision score from a mean of 0.61 ± 0.17 (SD) to 0.41 ± 0.25 in comparison to the initial classification with threshold selection based on F‐score. The use of a secondary, contextual classification with thresholding selection accounting for heterogeneity of precision reduced it further still, to 0.16 ± 0.13, and was significantly lower than the initial classification in all but one species. Mean average precision scores increased, from 0.66 ± 0.4 for the initial classification, to 0.95 ± 0.19, a significant improvement for all species. We recommend assessing—and if necessary correcting for—heterogeneity of precision error when using automated classification on acoustic data to quantify species presence as a function of an environmental, spatial or temporal predictor variable.
format Article
id doaj-art-aade5be4e7ba4818b7ac8f3f9da05a6a
institution DOAJ
issn 2041-210X
language English
publishDate 2022-11-01
publisher Wiley
record_format Article
series Methods in Ecology and Evolution
spelling doaj-art-aade5be4e7ba4818b7ac8f3f9da05a6a2025-08-20T03:02:28ZengWileyMethods in Ecology and Evolution2041-210X2022-11-0113112559257110.1111/2041-210X.13967Detecting and reducing heterogeneity of error in acoustic classificationOliver C. Metcalf0Jos Barlow1Yves Bas2Erika Berenguer3Christian Devenish4Filipe França5Stuart Marsden6Charlotte Smith7Alexander C. Lees8Division of Biology and Conservation Ecology, Department of Natural Sciences Manchester Metropolitan University Manchester UKLancaster Environment Centre Lancaster University Lancaster UKCESCO, MNHN, CNRS, Sorbonne Univ Paris FranceCESCO, MNHN, CNRS, Sorbonne Univ Paris FranceDivision of Biology and Conservation Ecology, Department of Natural Sciences Manchester Metropolitan University Manchester UKSchool of Biological Sciences University of Bristol Bristol UKDivision of Biology and Conservation Ecology, Department of Natural Sciences Manchester Metropolitan University Manchester UKCESCO, MNHN, CNRS, Sorbonne Univ Paris FranceDivision of Biology and Conservation Ecology, Department of Natural Sciences Manchester Metropolitan University Manchester UKAbstract Passive acoustic monitoring can be an effective method for monitoring species, allowing the assembly of large audio datasets, removing logistical constraints in data collection and reducing anthropogenic monitoring disturbances. However, the analysis of large acoustic datasets is challenging and fully automated machine learning processes are rarely developed or implemented in ecological field studies. One of the greatest uncertainties hindering the development of these methods is spatial generalisability—can an algorithm trained on data from one place be used elsewhere? We demonstrate that heterogeneity of error across space is a problem that could go undetected using common classification accuracy metrics. Second, we develop a method to assess the extent of heterogeneity of error in a random forest classification model for six Amazonian bird species. Finally, we propose two complementary ways to reduce heterogeneity of error, by (i) accounting for it in the thresholding process and (ii) using a secondary classifier that uses contextual data. We found that using a thresholding approach that accounted for heterogeneity of precision error reduced the coefficient of variation of the precision score from a mean of 0.61 ± 0.17 (SD) to 0.41 ± 0.25 in comparison to the initial classification with threshold selection based on F‐score. The use of a secondary, contextual classification with thresholding selection accounting for heterogeneity of precision reduced it further still, to 0.16 ± 0.13, and was significantly lower than the initial classification in all but one species. Mean average precision scores increased, from 0.66 ± 0.4 for the initial classification, to 0.95 ± 0.19, a significant improvement for all species. We recommend assessing—and if necessary correcting for—heterogeneity of precision error when using automated classification on acoustic data to quantify species presence as a function of an environmental, spatial or temporal predictor variable.https://doi.org/10.1111/2041-210X.13967automated signal recognitionautonomous recording unitbioacousticsecoacousticsmachine‐learning
spellingShingle Oliver C. Metcalf
Jos Barlow
Yves Bas
Erika Berenguer
Christian Devenish
Filipe França
Stuart Marsden
Charlotte Smith
Alexander C. Lees
Detecting and reducing heterogeneity of error in acoustic classification
Methods in Ecology and Evolution
automated signal recognition
autonomous recording unit
bioacoustics
ecoacoustics
machine‐learning
title Detecting and reducing heterogeneity of error in acoustic classification
title_full Detecting and reducing heterogeneity of error in acoustic classification
title_fullStr Detecting and reducing heterogeneity of error in acoustic classification
title_full_unstemmed Detecting and reducing heterogeneity of error in acoustic classification
title_short Detecting and reducing heterogeneity of error in acoustic classification
title_sort detecting and reducing heterogeneity of error in acoustic classification
topic automated signal recognition
autonomous recording unit
bioacoustics
ecoacoustics
machine‐learning
url https://doi.org/10.1111/2041-210X.13967
work_keys_str_mv AT olivercmetcalf detectingandreducingheterogeneityoferrorinacousticclassification
AT josbarlow detectingandreducingheterogeneityoferrorinacousticclassification
AT yvesbas detectingandreducingheterogeneityoferrorinacousticclassification
AT erikaberenguer detectingandreducingheterogeneityoferrorinacousticclassification
AT christiandevenish detectingandreducingheterogeneityoferrorinacousticclassification
AT filipefranca detectingandreducingheterogeneityoferrorinacousticclassification
AT stuartmarsden detectingandreducingheterogeneityoferrorinacousticclassification
AT charlottesmith detectingandreducingheterogeneityoferrorinacousticclassification
AT alexanderclees detectingandreducingheterogeneityoferrorinacousticclassification