I'm trying to solve a fitting problem with the Matlab Neural Network Toolbox. I must say i am a beginner with this tool and i would really appreciate your help.
I have a large data set, consisting of 17 elements with more than 30000 samples which i use as an Input and a Target data set. I use the GUI in NN Toolbox to load the data and perform the training. The results after training seem to be good, as i get a relatively small error. However when i load the second data set which i want to use for the simulation my error gets really high. I read about generalization problems and over fitting issues but I'm not sure how to deal with these and how to change my code so that i improve my simulation performance.
Furthermore I'm not sure about the setup of my network - number of hidden neurons, the percentages of Tr./Val./Test. etc?
I would be really thankful if anyone could help me out here. Thanks & best regards