中国可持续发展水平区域差异的人工神经网络判定

THE ANALYSIS ON REGIONAL DIFFERENTIATIONS OF SUSTAINABLE BY USING ARTIFICIAL NEURAL NETWORDS

本文在对目前区域可持续发展水平度量方法进行分析的基础上 ,提出以人工神经网络 (ANN)作为区分中国可持续发展水平区域差异的模型工具。构建了ANN模型分析中较为先进的自组织特征映射网络 (Self-Organiz ingMaps,SOFM) ,并以中国 31个省市 (自治区 ) 1996年的社会、经济、资源和环境状况作为待分样本 ,用SOFM进行了可持续发展区域差异的判定。网络运行结果表明 ,1996年中国可持续发展水平的区域差异可分成 5类 ,SOFM分类结果与专家的判断基本近似。可见 。

Artificial neural networks, originally inspired by their biological namesakes, are composed of many simple intercommunicating elements, or neurons, working in parallel to solve a problem. What makes them exciting is the fact that once a network has been set up, it can learn in self-organizing way that seems to mimic simple biological nervous systems. Because neural networks can be trained to respond in parallel to the inputs presented to them, they often are much faster than more conventional methods.In this paper, a soundly trained self-organizing map (SOFM), developed by Teuvo Kohonen in 1981, is employed to measure the sustainability of regional socioeconomic system. Without assuming parametric relationship, the neural network directly maps the patterns of socioeconomic system. When the neural network is trained appropriately, it classifies the data sets.The run results of SOFM show that 31provinces(cities) or autonomous regions are classified into 5 groups, which are in of accord with experts. The results also indicate that ANN is an alternative approach of assessing regional sustainability.