摘要
在传统小脑模型的基础上,利用其特有的寻址方式,引入模糊逻辑的思想,采用模糊隶属度函数作为接受域函数,提出了一种模糊小脑模型神经网络。重点研究了接受域函数的映射规律、隶属度函数及其参数的选取规律和学习算法。仿真结果表明,该模糊模型具有良好的泛化能力和逼近精度,且可获得连续性强且有解析微分输出的复杂函数。
Based on the traditional cerebellum model, utilizing the special addressing schemes, and introducing the fuzzy logic idea, a general neural network with cerebellum model was proposed in which the fuzzy membership functions were used as the receptive domain functions. The mapping law of receptive domain functions, the selection law of membership function and its parameters, and the learning algorithm were emphatically studied. The simulation results show that the fuzzy model has good generalizing capability and approximate accuracy, and also can get the complicated functions with strong continuity and numerical derivative output.
出处
《山东科技大学学报（自然科学版）》
CAS
2009年第1期70-73,共4页
Journal of Shandong University of Science and Technology(Natural Science)
基金
山东省优秀中青年科学家科研奖励基金项目(2006BS01018)
关键词
小脑模型
模糊逻辑
映射规律
学习算法
cerebellum model
fuzzy logic
mapping law
learning algorithm
