基于GA优化BP网络输入的人力资本预测研究

人力资本是决定地区经济增长的重要因素之一,其形成受多种因素的综合影响。利用BP神经网络进行预测时,因人力资本形成因素众多,各因素之间信息独立性差,交叉重叠的信息使得神经网络的结构过于庞大,极大地影响了网络的性能。本文采用遗传算法(GA)对人力资本形成因素进行优化,化选出与人力资本关系最密切、反应最敏感的少数指标,再利用优化后的指标对我国31个地区的人力资本进行预测,有效地降低了网络输入维数,提高了网络精度。

Study on Human Capital Forecast Based on GA Optimizing BP Network Inputs

Human capital is one of the most important factors in deciding economy growth of one region,whose formation is affected by many factors.When we use BP network to forecast,the size of ANN is so big that it obviously reduces the mapping capacity of network because of many less independent factors and overlap information.In this paper we use GA to optimize the factors which form the human capital and choose a few factors which are the most sensitive and have compact relation to human capital.Next we use these chosen indexes to forecast the human capital in 31 regions in China.It is proved that the dimensions of network input are reduced effectively and at the same time we improve the precision of network by the method.