The effectiveness of artificial neural networks in modelling the nutritional ecology of a blowfly species

• Autor(a) principal: Watts, Michael J.
• Data de Publicação: 2012
• Outros Autores: Bianconi, Andre [UNESP], Serapiao, Adriane Beatriz S. [UNESP], Govone, Jose S. [UNESP], Von Zuben, Claudio J. [UNESP]
• Tipo de documento: Capítulo de livro
• Idioma: eng
• Título da fonte: Repositório Institucional da UNESP
• Texto Completo: http://hdl.handle.net/11449/227500
• topic: Larval phase; Life history; Neural algorithms; Pupal mass; Regression models

The larval phase of most blowfly species is considered a critical developmental period in which intense limitation of feeding resources frequently occurs. Furthermore, such a period is characterised by complex ecological processes occurring at both individual and population levels. These processes have been analysed by means of traditional statistical techniques such as simple and multiple linear regression models. Nonetheless, it has been suggested that some important explanatory variables could well introduce non-linearity into the modelling of the nutritional ecology of blowflies. In this context, dynamic aspects of the life history of blowflies could be clarified and detailed by the deployment of machine learning approaches such as artificial neural networks (ANNs), which are mathematical tools widely applied to the resolution of complex problems. A distinguishing feature of neural network models is that their effective implementation is not precluded by the theoretical distribution of the data used. Therefore, the principal aim of this investigation was to use neural network models (namely multi-layer perceptrons and fuzzy neural networks) in order to ascertain whether these tools would be able to outperform a general quadratic model (that is, a second-order regression model with three predictor variables) in predicting pupal weight values (outputs) of experimental populations of Chrysomya megacephala (F.) (Diptera: Calliphoridae), using initial larval density (number of larvae), amount of available food, and pupal size as input variables. These input variables may have generated non-linear variation in the output values, and fuzzy neural networks provided more accurate outcomes than the general quadratic model (i.e. the statistical model). The superiority of fuzzy neural networks over a regression-based statistical method does represent an important fact, because more accurate models may well clarify several intricate aspects regarding the nutritional ecology of blowflies. Additionally, the extraction of fuzzy rules from the fuzzy neural networks provided an easily comprehensible way of describing what the networks had learnt. © 2012 by Nova Science Publishers, Inc. All rights reserved.