机构地区: 广西大学数学与信息科学学院
出 处: 《广西大学学报(自然科学版)》 2009年第5期635-639,共5页
摘 要: 预测高等教育投资供给规模,对于制定高等教育发展规划,确定高等教育发展规模十分必要。高等教育投资供给是一个非线性系统,而神经网络对非线性系统处理效果较好。为了改善预测性能,将神经网络训练算法进行改进。论文分析研究了L-M算法原理,将其用于高教投资供给规模的预测中,并对整个预测过程进行优化。实验结果表明,基于优化L-M算法的高等教育投资供给规模预测模型收敛速度快,泛化能力更优。 It is necessary to predict the investment supply scale to formulate the development program and determine the scale for higher education of China. Higher education investment supply is a non-linear system, and a good result can be achieved with a neural network. In order to increase the prediction performance ,the training algorithm of neural network was improved. The principle of the Levengerg-Marquardt(L-M) algorithm was studied and analyzed in this paper. The L-M algorithm Was used to predict the investment supply scale of higher education, and the entire prediction process was optimized. Experimental results indicated that the L-M optimization algorithm based investment supply scale prediction model in higher education showed faster converging speed and higer generalization ability.