自联想神经网络
Posted pprotector
tags:
篇首语:本文由小常识网(cha138.com)小编为大家整理,主要介绍了自联想神经网络相关的知识,希望对你有一定的参考价值。
第一篇博文,先贴代码慢慢改……
%------------------------AANN by NNtoolbox---------------------------------- tran_fun_type ={‘logsig‘;‘purelin‘;‘logsig‘;‘purelin‘}; bn_l = 1;% if bn_l<10 lh = (2+2*bn_l); else lh = bn_l; end net = feedforwardnet([lh,bn_l,lh]); %net.trainParam.epochs=1000% for i=1:size(tran_fun_type,1) net.layers{i}.transferFcn = tran_fun_type{i}; end net = train(net,data_1,data_1); input = mapminmax(‘apply‘,data_1,net.inputs{1}.processSettings{1}); L1 = logsig(net.IW{1,1} * input + net.B{1}); L2 = purelin( net.LW{2,1} * L1 + net.B{2}); L3 = logsig(net.LW{3,2} * L2 + net.B{3}); L4 = purelin(net.LW{4,3} * L3 + net.B{4}); OUT = mapminmax(‘reverse‘,L4,net.outputs{4}.processSettings{1}); data_resc1_1 = sim(net,data_1); data_resc1 = mapminmax(‘reverse‘,data_resc1_1,traM) dt = mapminmax(‘reverse‘,data_1,traM); for i = 1:size(data_1,1) fig_hdl = figure(i); y = data_1(i,:)‘; x = data_resc1_1(i,:)‘; FIT_CUR = polyfit(x,y,1); r_ = corrcoef(x,y); y_ = polyval(FIT_CUR,x); plot(x,y,‘k+‘,x,y_,‘r‘); title(strcat(‘R:‘,num2str(r_(2,1)),‘ a:‘,num2str(FIT_CUR(1)),‘ b:‘, num2str(FIT_CUR(2)) )); saveas(fig_hdl,strcat(‘AANNfit_cur‘,num2str(i)),‘jpg‘); end %--------------------------NLPCA toolbox------------------------------------------ [pc net network] = nlpca(data_1, bn_l); data_resc2_1 = nlpca_get_data(net,pc); for i = 1:size(data_1,1) fig_hdl = figure(i); y = data_1(i,:)‘; x = data_resc2_1(i,:)‘; FIT_CUR = polyfit(x,y,1); r_ = corrcoef(x,y); y_ = polyval(FIT_CUR,x); plot(x,y,‘k+‘,x,y_,‘r‘); title(strcat(‘R:‘,num2str(r_(2,1)),‘ a:‘,num2str(FIT_CUR(1)),‘ b:‘, num2str(FIT_CUR(2)) )); saveas(fig_hdl,strcat(‘NLPCAfit_cur‘,num2str(i)),‘jpg‘); end %--------------------------PCA ------------------------------------------ [coeff,score,latent,tsquared,explained,mu] = pca(data_1‘) tar=bn_l; data_resc3_1 = score(:,1:tar)*coeff(:,1:tar)‘ + repmat(mu,size(data_1,2),1); for i = 1:size(data_1,1) fig_hdl = figure(i); y = data_1(i,:)‘; x = data_resc3_1(:,i); FIT_CUR = polyfit(x,y,1); r_ = corrcoef(x,y); y_ = polyval(FIT_CUR,x); plot(x,y,‘k+‘,x,y_,‘r‘); title(strcat(‘R:‘,num2str(r_(2,1)),‘ a:‘,num2str(FIT_CUR(1)),‘ b:‘, num2str(FIT_CUR(2)) )); saveas(fig_hdl,strcat(‘PCAfit_cur‘,num2str(i)),‘jpg‘); end
以上是关于自联想神经网络的主要内容,如果未能解决你的问题,请参考以下文章