自组织竞争神经网络及其在社会经济区划中的应用

在MATLAB6.5软件的支持下,利用人工神经网络中"自组织竞争神经网络"的原理及聚类功能,选取耕地面积、人均耕地、农业总产值、乡镇企业产值、财政收入、农民人均收入等6项社会经济相关指标,较为客观地将宝兴县11个乡(镇)划分为3类社会经济区域,并对其分类结果进行了分析.结果表明,该网络能很好地反映并提取样本间复杂的信息,实现科学合理的社会经济区划.相比传统的系统聚类法,自组织竞争神经网络由于在运行过程中人为主观因素介入少,因此分类结果具有更大的可靠性,适于在社会经济分区中应用.

Self-organized Competing Neural Network and Its Application in Compartmentalizing Social Economy Areas

With the support of software MATLAB6.5,the theory and clustering function of self-organized competing neural network in Artificial NN was used to this study.The 6 social economy indexes were selected,which were cultivated land acreage,cultivated land acreage/ person,agricultural total production value,village business enterprise production value,public finance income and private income/ peasant.The 11 towns in Baoxing County were compartmentalized objectively into three social economy areas,and the compartmentalized results were analyzed.The results indicated that such network could reflect and pick up the complicated information among samples very well and then realize reasonable social economy area compartmentalization.Compared with traditional system clustering method,subjective factors in the running process of self-organized competing neural network are less,so its compartmentalized result is more reliable,and this network is suitable for application in social economy area compartmentalization.