why the number of epoches is low
1 view (last 30 days)
Show older comments
i am training the feedforward back propagation neural network using nntool in matlab with input vector of 12*304 and target vector of 1*304. Here is the list of parameters that I have used 2 hidden layers transfer function of (layer1,layer2,outputlayer)= logsig,logsig,tansig number of neurons(hiddenlayer1, hiddenlayer2)= 8,8
i tried many times. my epoches was less than 8. is this show a problem in my structure? in addition the error for the validation curveincreases with increasing epoch. is this show a wrong in nn structure?
0 Comments
Accepted Answer
Greg Heath
on 5 Feb 2015
The default structure is an I-H-O = 12-10-1 tansig/purelin fitnet. Standardize input and target with zscore. Try that configuration 10 times (10 different RNG states for data division and initial weights). If unsuccessful, systematically increase H. If validation stopping keeps occurring, try
net.divideFcn = '' % to get 100/0/0 data division.
Alternatively, since the default is
>> max_fail = net.trainParam.max_fail
max_fail = 6
Reset maxfail to a higher number
net.trainParam.max_fail = inf;
Hope this helps.
Thank you for formally accepting my answer
Greg
0 Comments
More Answers (0)
See Also
Categories
Find more on Image Data Workflows in Help Center and File Exchange
Products
Community Treasure Hunt
Find the treasures in MATLAB Central and discover how the community can help you!
Start Hunting!