In:
PLOS Computational Biology, Public Library of Science (PLoS), Vol. 19, No. 4 ( 2023-4-13), p. e1010325-
Abstract:
Despite the accumulation of data and studies, deciphering animal vocal communication remains challenging. In most cases, researchers must deal with the sparse recordings composing Small, Unbalanced, Noisy, but Genuine (SUNG) datasets. SUNG datasets are characterized by a limited number of recordings, most often noisy, and unbalanced in number between the individuals or categories of vocalizations. SUNG datasets therefore offer a valuable but inevitably distorted vision of communication systems. Adopting the best practices in their analysis is essential to effectively extract the available information and draw reliable conclusions. Here we show that the most recent advances in machine learning applied to a SUNG dataset succeed in unraveling the complex vocal repertoire of the bonobo, and we propose a workflow that can be effective with other animal species. We implement acoustic parameterization in three feature spaces and run a Supervised Uniform Manifold Approximation and Projection (S-UMAP) to evaluate how call types and individual signatures cluster in the bonobo acoustic space. We then implement three classification algorithms (Support Vector Machine, xgboost, neural networks) and their combination to explore the structure and variability of bonobo calls, as well as the robustness of the individual signature they encode. We underscore how classification performance is affected by the feature set and identify the most informative features. In addition, we highlight the need to address data leakage in the evaluation of classification performance to avoid misleading interpretations. Our results lead to identifying several practical approaches that are generalizable to any other animal communication system. To improve the reliability and replicability of vocal communication studies with SUNG datasets, we thus recommend: i) comparing several acoustic parameterizations; ii) visualizing the dataset with supervised UMAP to examine the species acoustic space; iii) adopting Support Vector Machines as the baseline classification approach; iv) explicitly evaluating data leakage and possibly implementing a mitigation strategy.
Type of Medium:
Online Resource
ISSN:
1553-7358
DOI:
10.1371/journal.pcbi.1010325
DOI:
10.1371/journal.pcbi.1010325.g001
DOI:
10.1371/journal.pcbi.1010325.g002
DOI:
10.1371/journal.pcbi.1010325.g003
DOI:
10.1371/journal.pcbi.1010325.g004
DOI:
10.1371/journal.pcbi.1010325.g005
DOI:
10.1371/journal.pcbi.1010325.g006
DOI:
10.1371/journal.pcbi.1010325.g007
DOI:
10.1371/journal.pcbi.1010325.g008
DOI:
10.1371/journal.pcbi.1010325.g009
DOI:
10.1371/journal.pcbi.1010325.g010
DOI:
10.1371/journal.pcbi.1010325.g011
DOI:
10.1371/journal.pcbi.1010325.g012
DOI:
10.1371/journal.pcbi.1010325.g013
DOI:
10.1371/journal.pcbi.1010325.g014
DOI:
10.1371/journal.pcbi.1010325.g015
DOI:
10.1371/journal.pcbi.1010325.t001
DOI:
10.1371/journal.pcbi.1010325.t002
DOI:
10.1371/journal.pcbi.1010325.t003
DOI:
10.1371/journal.pcbi.1010325.s001
DOI:
10.1371/journal.pcbi.1010325.s002
DOI:
10.1371/journal.pcbi.1010325.s003
DOI:
10.1371/journal.pcbi.1010325.s004
DOI:
10.1371/journal.pcbi.1010325.s005
DOI:
10.1371/journal.pcbi.1010325.s006
DOI:
10.1371/journal.pcbi.1010325.s007
DOI:
10.1371/journal.pcbi.1010325.s008
DOI:
10.1371/journal.pcbi.1010325.s009
DOI:
10.1371/journal.pcbi.1010325.s010
DOI:
10.1371/journal.pcbi.1010325.r001
DOI:
10.1371/journal.pcbi.1010325.r002
DOI:
10.1371/journal.pcbi.1010325.r003
DOI:
10.1371/journal.pcbi.1010325.r004
DOI:
10.1371/journal.pcbi.1010325.r005
DOI:
10.1371/journal.pcbi.1010325.r006
Language:
English
Publisher:
Public Library of Science (PLoS)
Publication Date:
2023
detail.hit.zdb_id:
2193340-6