If there is less values, you need to use root mean square errors. I guess that it is better.

use other measures, also check this:

Srivastava, N., Hinton, G. E., Krizhevsky, A., Sutskever, I., & Salakhutdinov, R. (2014). Dropout: a simple way to prevent neural networks from overfitting. Journal of Machine Learning Research, 15(1), 1929-1958.

 Hi monther,

I have 40 number of random data including training and testing. As the process is time consuming so it is difficult to increase number of data sets. The neural network is of size 4-n-1 where n is number of neurons in the hidden layer.

How to overcome overfitting in ANN?

Hi Monther,

I use Matlab 14a 'nn wizard'  for ANN architecture