作者Masashi0620 (三上真史)
看板MATLAB
标题[问题] 类神经网路 Bug 错误...
时间Tue Aug 29 17:14:22 2006
各位大大...
小弟刚刚初试啼声创建了一个倒传递类神经程式...
function test1
P = [0 1 2 3 4 5 6 7 8 9 10];
T = [0 1 2 3 4 3 2 1 2 3 4];
net = newff(minmax(P),[5 1],{'tansig' 'purelin'},'traingd');
net.trainParam.show = 100;
net.trainParam.lr = 0.05;
net.trainParam.epochs = 500;
[net.tr] = train(net,P,T);
去Run 确实可以Run 也有图形出来...
可是还是出现下面的讯息 好像是说有错误...= =
有哪位大大可以帮我解释一下吗???
??? Error using ==> network.subsasgn
Reference to non-existent field.
Error in ==> test1 at 10
[net.tr] = train(net,P,T);
TRAINGD-calcgrad, Epoch 0/500, MSE 12.729/0, Gradient 14.86/1e-010
TRAINGD-calcgrad, Epoch 100/500, MSE 0.082174/0, Gradient 0.133331/1e-010
TRAINGD-calcgrad, Epoch 200/500, MSE 0.044553/0, Gradient 0.0500553/1e-010
TRAINGD-calcgrad, Epoch 300/500, MSE 0.0383653/0, Gradient 0.0234007/1e-010
TRAINGD-calcgrad, Epoch 400/500, MSE 0.0366584/0, Gradient 0.0146977/1e-010
TRAINGD-calcgrad, Epoch 500/500, MSE 0.0358333/0, Gradient 0.0113496/1e-010
TRAINGD, Maximum epoch reached, performance goal was not met.
??? Error using ==> network.subsasgn
Reference to non-existent field.
Error in ==> test1 at 10
[net.tr] = train(net,P,T);
--
※ 发信站: 批踢踢实业坊(ptt.cc)
◆ From: 203.73.164.48
1F：推 asingular:"net"这个structure里面没有tr这个元素吧 09/01 11:17
2F：→ asingular:把训练结果存回net，或是指定给一个新的net也可以 09/01 11:19