i created a copy of trainlm algorithm in neural network to shivlm.m with the same code of the trainlm.m. when i tried to call shivlm i get errors. can anyone help me in this regard?

the error i get when i copy the file trainlm.m and renamed the file as shivlm.

 Undefined function 'nn_train_feedback' for input
arguments of type 'struct'.

Error in shivlm>train_network (line 206) nn_train_feedback('start',archNet,status);

Error in shivlm (line 64) [out1,out2] = train_network(varargin{2:end});

Error in network/train (line 329) [calcNet,tr] = trainFcn('apply',net,rawData,calcLib,calcNet,tr);

Error in cmpare (line 22) [net,pr]=train(net,Input',Target');

Answer 1

[Walter Roberson]

trainlm probably uses private functions that you have not copied. You might need to copy the entire class directory.

[Dhandapani.S]

thank you Walter. i found C:\Program Files (x86)\MATLAB\R2010a\toolbox\nnet\nnet\nntrain . should i copy the nnet directory to the mydocuments\matlab folder which is the current folder i am using?

[Dhandapani.S]

i find two trainlm.m .one using the open trainlm.m in the command windows of matlab. other in the c:\Program Files (x86)\MATLAB\R2010a\toolbox\nnet\nnet\nntrain. both are different files.

[Dhandapani.S]

i wanted to do some modification in the trainlm algorithm in the matlab and try to improve the performance. please help what are the files and pvt folders to be copied so that i can try with the optimization of trainlm.

[Walter Roberson]

Probably. You might need to copy even more than that, other subdirectories in addition to nntrain .

[Dhandapani.S]

can you please tell me the files or folders i wanted to copy to my working folder. because the files i cannot edit because it is read only

[Walter Roberson]

Which MATLAB release are you using?

[Walter Roberson]

Later postings indicate the user is using R2012b. This matters because the Neural Network toolbox was completely rewritten between then and now, so the functions the user is having difficulty with do not exist in the current version.

[Walter Roberson]

You need the private subdirectory, which is at

fullfile(toolboxdir('nnet'),'nnet','nntrain', 'private')