例如：我需要這個神經(jīng)網(wǎng)絡(luò)輸出值符合我想要的范圍內(nèi)，我又不想一直點運行。

此時可以使用如下構(gòu)架：

while 1==1

if 要求
break
end

end

例子：

clc
clear
close all 


while 1==1
%%

load data_xy.mat
data_y = data(1:50000,1:1);
data_x = data(1:50000,2:2);
%構(gòu)建
train_x = [data_x]';%數(shù)據(jù)
train_y = [data_y]';%標簽

%構(gòu)建網(wǎng)絡(luò)
net = newff(train_x,train_y, [10], { 'logsig' 'purelin' } , 'traingdx' , 'learngdm') ;%輸入數(shù)據(jù)為 特征數(shù)*數(shù)據(jù)個數(shù)，輸出為 類別向量*數(shù)據(jù)個數(shù)
net.trainParam.showWindow = 0;% 不顯示窗口
net.trainparam.epochs = 5000;%允許最大訓(xùn)練步數(shù)500步
net.trainparam.goal = 0.00000001 ;%訓(xùn)練目標最小誤差0.01
net.trainParam.lr = 0.0000001 ;%學(xué)習速率0.05
%% 開始訓(xùn)練
net = train( net, train_x,train_y);
close all 
%% 仿真測試
predict_y = sim( net,train_x) ;

SST = sum((train_y-mean(train_y)).^2);
SSR = sum((predict_y'-mean(train_y)).^2);

R2 = SSR./SST

if 0.998 <R2&&R2< 1.002
    save good2.mat
    x = 1:50000;
    plot(x,train_y,'r',x,predict_y,'g');
    break
end
end