资源描述
单击此处编辑母版标题样式,单击此处编辑母版文本样式,第二级,第三级,第四级,第五级,*,*,单击此处编辑母版标题样式,单击此处编辑母版文本样式,第二级,第三级,第四级,第五级,*,单击此处编辑母版标题样式,单击此处编辑母版文本样式,第二级,第三级,第四级,第五级,/10/29,.,*,数学建模与数学实验,神经网络,数学建模与数学实验 神经网络,目的,内容,学习神经网络的基本原理与方法。,1,、人工神经元数学模型,4,、,BP,神经网络应用,3,、,BP,神经网络,Matlab,工具箱函数,2,、,BP,神经网络,目的内容学习神经网络的基本原理与方法。1、人工神经元数学模型,神经网络算法课件,一、人工神经元数学模型,一、人工神经元数学模型,神经网络算法课件,神经网络算法课件,神经网络算法课件,BP,神经网络的拓扑结构如图所示。,1.BP,神经网络结构:,BP神经网络的拓扑结构如图所示。1.BP神经网络结构:,神经网络算法课件,2,BP,神经网络学习算法及流程,以三层,BP,神经网络为例,它的训练过程包括以下几个步骤:,2BP神经网络学习算法及流程以三层BP神经网络为例,它的训,神经网络算法课件,神经网络算法课件,BP,神经网络的流程图:,BP神经网络的流程图:,三、,BP,神经网络,Matlab,工具箱函数,三、BP神经网络Matlab工具箱函数,net=newff(PR,S1,S2,SN,TF1,TF2,TFN,BTF,BLF,PF),net=newff(PR,S1,S2,SN,T,神经网络算法课件,神经网络算法课件,神经网络算法课件,神经网络算法课件,神经网络算法课件,神经网络算法课件,网络经过,177,次训练后,虽然网络的性能还没有达到,0,,但是输出的均方误差已经很小了,,MSE,2.95307e-006,,误差曲线如图,1,所示。为更直观地理解网络输出与目标向量之间的关系,见图,2,所示。,plot(P,T,-,P,Y,o),图,1 BP,神经网络训练误差曲线图 图,2,训练后,BP,神经网络仿真图,To Matlab exp12_4_1.m,网络经过177次训练后,虽然网络的性能还没有达到0,,神经网络算法课件,神经网络算法课件,神经网络算法课件,神经网络算法课件,神经网络算法课件,神经网络算法课件,BP,神经网络测试结果图,To Matlab exp12_4_2.m,BP神经网络测试结果图 To Matlab exp12_4_,神经网络算法课件,神经网络算法课件,神经网络算法课件,神经网络算法课件,神经网络算法课件,神经网络算法课件,神经网络算法课件,神经网络算法课件,神经网络算法课件,To Matlab exp12_4_3.m,To Matlab exp12_4_3.m,神经网络算法课件,练习,1,、蠓虫分类问题,生物学家试图对两种蠓虫(,Af,与,Apf,)进行鉴别,依据的资料是触角和翅膀的长度,已经测得了,9,只,Af,和,6,只,Apf,的数据如下:,Af:(1.24,1.27),,,(1.36,1.74),,,(1.38,1.64),,,(1.38,1.82),,,(1.38,1.90),,,(1.40,1.70),,,(1.48,1.82),,,(1.54,1.82),,,(1.56,2.08),;,Apf:(1.14,1.82),,,(1.18,1.96),,,(1.20,1.86),,,(1.26,2.00),,,(1.28,2.00),,,(1.30,1.96).,(,i,)根据如上资料,如何制定一种方法,正确地区分两类蠓虫;,(,ii,)对触角和翼长分别为,(1.24,1.80),,,(1.28,1.84),与,(1.40,2.04),的三个标本,用所得到的方法加以识别;,(,iii,)设,Af,是宝贵的传粉益虫,,Apf,是某疾病的载体,是否应该修改分类方法,.,练习1、蠓虫分类问题,2,、人口预测下表是从,北京统计年鉴,中给出的,1980-2010,年的北京城近郊区户籍人口统计结果作为样本数据,,(1),建立人工神经网络模型,;(2),预测,2011,年的北京城近郊区户籍人口,.,2、人口预测下表是从北京统计年鉴中给出的 1980-2,数据处理后的样本数据:,样本用途,样本组数,输入一,输入二,输入三,输入四,输出,学,习,样,本,1,0.4984,0.5102,0.5213,0.534,0.5407,2,0.5102,0.5213,0.534,0.5407,0.5428,3,0.5213,0.534,0.5407,0.5428,0.553,4,0.534,0.5407,0.5428,0.553,0.5632,5,0.5407,0.5428,0.553,0.5632,0.5739,6,0.5428,0.553,0.5632,0.5739,0.5821,7,0.553,0.5632,0.5739,0.5821,0.592,8,0.5632,0.5739,0.5821,0.592,0.5987,9,0.5739,0.5821,0.592,0.5987,0.6043,10,0.5821,0.592,0.5987,0.6043,0.6095,11,0.592,0.5987,0.6043,0.6095,0.6161,12,0.5987,0.6043,0.6095,0.6161,0.6251,13,0.6043,0.6095,0.6161,0.6251,0.6318,14,0.6095,0.6161,0.6251,0.6318,0.6387,15,0.6161,0.6251,0.6318,0.6387,0.6462,16,0.6251,0.6318,0.6387,0.6462,0.6518,17,0.6318,0.6387,0.6462,0.6518,0.6589,18,0.6387,0.6462,0.6518,0.6589,0.6674,19,0.6462,0.6518,0.6589,0.6674,0.6786,20,0.6518,0.6589,0.6674,0.6786,0.6892,21,0.6589,0.6674,0.6786,0.6892,0.6988,22,0.6674,0.6786,0.6892,0.6988,0.7072,检,验,样,本,23,0.6786,0.6892,0.6988,0.7072,0.7132,24,0.6892,0.6988,0.7072,0.7132,0.7185,25,0.6988,0.7072,0.7132,0.7185,0.7309,26,0.7072,0.7132,0.7185,0.7309,0.7438,27,0.7132,0.7185,0.7309,0.7438,0.7496,数据处理后的样本数据:样本用途样本组数输入一输入二输入三输入,神经网络算法课件,/10/29,45,.,/10/2945.,
展开阅读全文