《BP神经网络的数据分类-MATLAB源代码》由会员分享，可在线阅读，更多相关《BP神经网络的数据分类-MATLAB源代码（8页珍藏版）》请在金锄头文库上搜索。
1、%清除空间clcclear all ; close all ;%训练数据预测数据提取以及归一化%下载四类数据load data1 c1load data2 c2load data3 c3load data4 c4%四个特征信号矩阵合成一个矩阵data ( 1:500 , : ) = data1 ( 1:500 , :) ;data ( 501:1000 , : ) = data2 ( 1:500 , : ) ;data ( 1001:1500 , : ) = data3 ( 1:500 , : ) ;data ( 1501:2000 , : ) = data4 ( 1:500 , : ) ;%
2、从1到2000间的随机排序k = rand ( 1 , 2000 ) ; m , n = sort ( k ) ; %m为数值，n为标号%输入输出数据input = data ( : , 2:25 ) ;output1 = data ( : , 1) ;%把输出从1维变到4维for i = 1 : 1 :2000switch output1( i ) case 1output( i , :) = 1 0 0 0 ;case 2 output( i , :) = 0 1 0 0 ;case 3output( i , :) = 0 0 1 0 ;case 4output( i , :) = 0 0
3、 0 1 ;end end%随机抽取1500个样本作为训练样本，500个样本作为预测样本input_train = input ( n( 1:1500 , : ) ) ;output_train = output ( n( 1:1500 , : ) ) ; input_test = input ( n( 1501:2000 , : ) ) ; output_test = output ( n( 1501:2000 , : ) ) ;%输入输出数据归一化 inputn , inputps = mapminmax ( input_train ) ;%网络结构初始化innum = 24 ; %输入层
4、midnum = 25 ; %隐含层outnum = 4 ; %输出层%权值初始化w1 = rands ( midnum , innum ) ; b1 = rands ( midnum , 1 ) ; w2 = rands ( midnum , outnum ) ;b2 = rands ( outnum , 1) ;w2_1 = w2 ; w2_2 = w2_1 ;w1_1 = w1 ; w1_2 = w1_1 ;b1_1 = b1 ; b1_2 = b1_1 ;b2_1 = b2 ; b2_2 = b2_1 ;%学习速率xite = 0.1 ;alfa = 0.01 ;%网络训练for ii
5、 = 1:10E( ii ) = 0 ;for i = 1:1:1500 ;%网络预测输出x = inputn ( : , j ) ;%隐含层输出for j = 1:1:midnuml (j) = inputn ( : , i )*w1( j , : ) + b1 (j) ;lout (j) = 1/( 1 +exp( -1(j) ) ) ;end%输出层输出yn = w2 * lout + b2 ;%权值阈值修正%计算权值变化率dw2 = e * lout ;db2 = e ;for j = 1:1:midnum S= 1/(1 + exp ( -l(j) ) ) ;Fl (j) = S *
6、 ( 1- S) ;endfor k = 1:1:innumfor j = 1:1:midnumdw1( k, j ) = Fl (j) * x (k) *( e(1)*w2( j,1) + e(2)*w2( j,2) + e(3)*w2( j,3) + e(4)*w2( j,4) ) ;db1( j ) = Fl (j) * *( e(1)*w2( j,1) + e(2)*w2( j,2) + e(3)*w2( j,3) + e(4)*w2( j,4) ) ;end endw1=w1_1+xite*dw1;b1=b1_1+xite*db1;w2=w2_1+xite*dw2;b2=b2_1+x
7、ite*db2; w1_2=w1_1;w1_1=w1;w2_2=w2_1;w2_1=w2;b1_2=b1_1;b1_1=b1; b2_2=b2_1;b2_1=b2;endend %语音特征信号分类input_test = mapminmax ( apply , input_test , inputps );for ii = 1:1for i = 1:500%隐含层输出for j = 1:1:midnuml (j) = input_test ( : , i ) * w1( j , : ) + b1(j) ; lout ( j ) = 1/ ( 1 + exp( -l(j) ) ) ;endfor
8、e( :,i ) = w2 * lout + b2 ;endend%结果分析%根据网络输出找出数据属于哪类for i = 1:500output_fore (i) = find ( fore (:,i) = =max (fore(:,i) ) ) ;end%BP网络预测输出error = output_fore - output1 ( n( 1501:2000) ) ;%画出分类图figure (1)plot ( output_fore , r ) hold onplot (output1( n (1501:2000) , b ) ;legend ( 预测语音类别 , 实际语音类别 )%画出误
9、差图figure (2)plot (error)title ( BP网络分类误差 , fontsize , 12 )xlabel ( 语音信号 , fontsize , 12 )ylabel ( 分类误差 , fontsize , 12 )%找出属于哪种类型for i = 1:500if error (i) = 0 b,c = max (output_test( :,i ) );switch ccase 1k(1) = k(1) + 1 ;case 2k(2) = k(2) + 1 ;case 3k(3) = k(3) + 1 ;case 4k(4) = k(4) + 1 ;endendend%找出每一类的个体总和kk = zeros ( 1,4 )for i = 1:500 b,c = max ( output_test( :,i) ) ;switch ccase 1kk(1) = kk(1) + 1 ;case 2kk(2) = kk(2) + 1 ;case 3kk(3) = kk(3) + 1 ;case 4kk(4) = kk(4) + 1 ;endend%正确率nightridio = ( kk - k )./ kk
