Machine learning techniques are becoming an important tool for studying animal vocal communication. The goat (Capra hircus) is a very social species, in which vocal communication and recognition are important. We tested the reliability of a Multi-Layer Perceptron (feed-forward ANN) to automate the process of classification of calls according to individual identity, group membership and maturation in this species. Vocalisations were obtained from 10 half-sibling (same father but different mothers) goat kids, belonging to 3 distinct social groups. We recorded 157 contact calls emitted during first week, and 164 additional calls recorded from the same individuals at 5 weeks. For each call, we measured 27 spectral and temporal acoustic parameters using a custom built program in the Praat software. For each classification task we built stratified 10-fold cross-validated neural networks. The input nodes corresponded to the acoustic parameters measured on each signal. ANNs were trained with the error-back-propagation algorithm. The number of hidden units was set to the number of attributes + classes. Each model was trained for 300 epochs (learning rate 0.2; momentum 0.2). To estimate a reliable error of the models, we repeated 10-fold cross-validation iterations 10 times and calculated the average predictive performance. The accuracy was 71.13 ± 1.16% for vocal individuality, 79.59 ± 0.75% for social group and 91.37 ± 0.76% for age of the vocalising animal . Our results demonstrate that ANNs are a powerful tool for studying vocal cues to individuality, group membership and maturation in contact calls. The performances we achieved were higher than those reported for similar classification tasks using classical statistical methods such as Discriminant Function Analysis. Further studies, investigating the reliability of these algorithms for the real-time classification of contact calls and comparing ANNs with other machine learning techniques are important to develop technology to remotely monitor the vocalisations of domestic livestock.

Artificial Neural Network approach for revealing individuality, group membership and age information in goat kid contact calls

FAVARO, LIVIO
First
;
2014-01-01

Abstract

Machine learning techniques are becoming an important tool for studying animal vocal communication. The goat (Capra hircus) is a very social species, in which vocal communication and recognition are important. We tested the reliability of a Multi-Layer Perceptron (feed-forward ANN) to automate the process of classification of calls according to individual identity, group membership and maturation in this species. Vocalisations were obtained from 10 half-sibling (same father but different mothers) goat kids, belonging to 3 distinct social groups. We recorded 157 contact calls emitted during first week, and 164 additional calls recorded from the same individuals at 5 weeks. For each call, we measured 27 spectral and temporal acoustic parameters using a custom built program in the Praat software. For each classification task we built stratified 10-fold cross-validated neural networks. The input nodes corresponded to the acoustic parameters measured on each signal. ANNs were trained with the error-back-propagation algorithm. The number of hidden units was set to the number of attributes + classes. Each model was trained for 300 epochs (learning rate 0.2; momentum 0.2). To estimate a reliable error of the models, we repeated 10-fold cross-validation iterations 10 times and calculated the average predictive performance. The accuracy was 71.13 ± 1.16% for vocal individuality, 79.59 ± 0.75% for social group and 91.37 ± 0.76% for age of the vocalising animal . Our results demonstrate that ANNs are a powerful tool for studying vocal cues to individuality, group membership and maturation in contact calls. The performances we achieved were higher than those reported for similar classification tasks using classical statistical methods such as Discriminant Function Analysis. Further studies, investigating the reliability of these algorithms for the real-time classification of contact calls and comparing ANNs with other machine learning techniques are important to develop technology to remotely monitor the vocalisations of domestic livestock.
2014
100
4
782
789
http://www.acta-acustica-united-with-acustica.com
Bioacoustics; machine learning; vocal communication; Ungulates
Livio Favaro; Elodie F. Briefer; Alan G. McElligott
File in questo prodotto:
File Dimensione Formato  
Favaro_2014.pdf

Accesso riservato

Tipo di file: PDF EDITORIALE
Dimensione 472.16 kB
Formato Adobe PDF
472.16 kB Adobe PDF   Visualizza/Apri   Richiedi una copia
AAA_Favaro_et_al_2014.pdf

Open Access dal 01/08/2015

Tipo di file: POSTPRINT (VERSIONE FINALE DELL’AUTORE)
Dimensione 426.56 kB
Formato Adobe PDF
426.56 kB Adobe PDF Visualizza/Apri

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/2318/141921
Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus 20
  • ???jsp.display-item.citation.isi??? 20
social impact