文档下载 文档下载导航
设为首页 | 加入收藏
搜索 请输入内容:  
 导航当前位置: 文档下载 > 所有分类 > 自然科学 > 数学 > 1996-Lek-EM-Application of neural networks to modelling nonlinear relationships in ecology
侵权投诉

1996-Lek-EM-Application of neural networks to modelling nonlinear relationships in ecology

ELSEVIER Ecological Modelling 90 (1996) 39-52 Application of neural networks to modelling nonlinear relationships in ecology Sovan Lek a,*, Marc Delacoste b Philippe Baran b, Ioannis Dimopoulos a Jacques Lauga a, St6phane Aulagnier c a UMR 9964 Equipe de Biologie Quantitative, Universit~ Paul Sabatier, 118 route de Narbonne, 31062 Toulouse, France b Laboratoire d'lng~nierie Agronomique, Equipe Ichtyologie Appliqu~e, ENSAT, 145 avenue de Muret, 31076 Toulouse, France c lnstitut de Recherche sur les Grands Mammi3~res, B.P. 27, 31326 Castanet-Tolosan, France Received 2 November 1994; accepted 23 June 1995 Abstract Two predictive modelling principles are discussed: multiple regression (MR) and neural networks (NN). The MR principle of linear modelling often gives low performance when relationships between variables are nonlinear; this is often the case in ecology; some variables must therefore be transformed. Despite these manipulations, the results often remain disappointing: poor prediction, dependence of residuals on the variable to predict. On the other hand NN are nonlinear type models. They do not necessitate transformation of variables and can give better results. The application of these two techniques to a set of ecological data (study of the relationship between density of brown trout spawning sites (redds) and habitat characteristics), shows that NN are clearly more performant than MR ( R E = 0.96 v s R 2 = 0.47 o r R E = 0.72 in raw variables or nonlinear transformed variables). With the calculation power now currently available, NN are easy to implement and can thus be recommended for modelling of a number ecological processes. Keywords: Model comparison; Multiple regression; Neural networks; Nonlinear relationships; Trout 1. Introduction For a number of years, the growing use of the aquatic ecosystem has forced the administration to address important economic and ecological problems. Habitat models, either deterministic or stochastic, based on the relationships between environment variables and characteristics of fish populations (density and biomass, reproduction potential, growth, etc.) are excellent tools for the managers. Faush et * Corresponding author. E-mail: lek@cict.fr; Fax: (+33-61) 556196. SSDI 0304-3800(95)00142-5 al. (1988) counted 95 models of this type. Of these 95 models published between 1950 and 1985, 79 were constructed from linear or multiple linear regression or correlation. This procedure assumes a linear relationship between the variables, which is rather rare in ecology. Then, 11 were curvilinear functions (exponential or power), and 5 were polynomial functions (but these were built up with very few variables and observations). Since this review, linear models have still been generally used (Milner et al., 1985; Jowett, 1990; etc.). On the other hand, artificial neural network (NN) systems are known for their capacity to process 0304-3800/96/$15.00 Copyright © 1996 Elsevier Science B.V. All rights reserve

第1页

你可能喜欢

热门文档

相关文档

站点地图 | 文档上传 | 侵权投诉 | 手机版
本站所有资源均来自互联网,本站只负责收集和整理,均不承担任何法律责任,如有侵权等其它行为请联系我们.
文档下载 Copyright 2013 doc.guandang.net All Rights Reserved.  -  email