Anomaly detection in honey bees (Apis mellifera L.) acoustics, temperature and humidity seasonal patterns
Abstract
Colony Collapse Disorder (CCD) is a phenomenon related to the sudden disappearance of honey bees in managed colonies. Registered in the USA since 2006, the possible causes of CCD are climatic variations, stress caused by the hives transport, inappropriate management of colonies, incorrect use of chemical pesticides, malnutrition, diseases, pests, among others. In this sense, creative computational solutions can contribute to a better understanding of the bees health and welfare. Here, we apply machine learning models to detect anomalies in acoustic patterns of Africanized honey bees (Apis mellifera L.) and in seasonal temperature and humidity patterns of European beehives. Three predictive models were implemented: Gaussian Mixture Model (GMM), Extreme Learning Machine (ELM) and Support Vector Machine for a class (OC-SVM). We used datasets with temperature and humidity from two European hives located in Bournemouth (England) and Würtzburg (Germany), as well audio from one colony located in Fortaleza-CE (Brazil). For temperature and humidity, the best results were for seasonal anomalies, in which the ELM algorithm reached an average accuracy of 92.6%. Regarding audio, we highlight the GMM algorithm (average accuracy of 84.9%) for queenless state detection using acoustic patterns data.
References
Braga, A. R., Gomes, D. G., Freitas, B. M., and Cazier, J. A. (2020). A cluster-classificationmethod for accurate mining of seasonal honey bee patterns.Ecological Informatics,59:101107.
Brown, M. J., Dicks, L. V., Paxton, R. J., Baldock, K. C., Barron, A. B., Chauzat, M.-P.,Freitas, B. M., Goulson, D., Jepsen, S., Kremen, C., et al. (2016). A horizon scan offuture threats and opportunities for pollinators and pollination.PeerJ, 4:e2249.
Chen, Y., Zhou, X. S., and Huang, T. S. (2001). One-class svm for learning in imageretrieval. InProceedings 2001 International Conference on Image Processing (Cat.No. 01CH37205), volume 1, pages 34–37. IEEE.
Davidson, P., Steininger, M., Lautenschlager, F., Kobs, K., Krause, A., and Hotho, A.(2020). Anomaly detection in beehives using deep recurrent autoencoders.arXivpreprint arXiv:2003.04576
Howard, D., Duran, O., Hunter, G., and Stebel, K. (2013). Signal processing the acousticsof honeybees (apis mellifera) to identify the "queenless"state in hives.Proceedingsof the Institute of Acoustics, 35:290–297.
Huang, G.-B., Zhu, Q.-Y., and Siew, C.-K. (2006). Extreme learning machine: theory andapplications.Neurocomputing, 70(1-3):489–501.
Klein, A.-M., Vaissiere, B. E., Cane, J. H., Steffan-Dewenter, I., Cunningham, S. A.,Kremen, C., and Tscharntke, T. (2007).Importance of pollinators in changinglandscapes for world crops.Proceedings of the royal society B: biological sciences,274(1608):303–313.
Lautenbach, S., Seppelt, R., Liebscher, J., and Dormann, C. F. (2012). Spatial and tem-poral trends of global pollination benefit.PLoS one, 7(4):e35954.
Ollerton, J., Winfree, R., and Tarrant, S. (2011). How many flowering plants are pollina-ted by animals?Oikos, 120(3):321–326.
Pimentel, M. A., Clifton, D. A., Clifton, L., and Tarassenko, L. (2014). A review of noveltydetection.Signal Processing, 99:215–249.
Reynolds, D. (2009).Gaussian Mixture Models, pages 659–663. Springer US, Boston,MA.
Silva, D., Ícaro Rodrigues, Braga, A., Nobre, J., Freitas, B., and Gomes, D. (2020). Anautonomic, adaptive and high-precision statistical model to determine bee colonieswell-being scenarios. InAnais do XI Workshop de Computação Aplicada à Gestão doMeio Ambiente e Recursos Naturais, pages 31–40, Porto Alegre, RS, Brasil. SBC.
Sánchez-Bayo, F. and Wyckhuys, K. A. (2019). Worldwide decline of the entomofauna:A review of its drivers.Biological Conservation, 232:8–27.
Zacepins, A., Brusbardis, V., Meitalovs, J., and Stalidzans, E. (2015). Challenges in thedevelopment of precision beekeeping.Biosystems Engineering, 130:60–71.
