As you knew, An Artificial Neural Network (ANN) is a mathematical model inspired by the biological behaviour of neurons and by the structure of the brain, which is used to solve a wide range of problems.

The main advantage of ANN models over the statistical methods is that the latter assume linear relationships and/or normal distribution, while reality is non-linear and non-normal. Thus the ANN model is capable to conform to the real world.

The ANN depends essentially on the exact information of the system under study, and the methods of training that must be used, as the algorithm of ANN during their training have the ability of identifying unnecessary data.

So that for your inquiry if you are training using real data and the behaviour of the resulted data satisfied the reality of the data, then the result can be considered as enough to simulate the model.

But in sometimes an ANN is not able to produce a method better than the statistic analysis applied to the spectroscopic techniques.

If you think answer help you, please recommend it.

Thanks Dr. Raheem Kh. Al-Sabur for your answer, but still ask , if i have ,for example, 10 cases can i generate 1000 cases to develop ANN model.

Dear Dr Gafel

As I mentioned before, ANN is good tool but not always the perfect selection.

To be able to get valuable results and able to use them in your model, your ANN generalisation must be satisfied your testing data. In ANN your data must be divided into two parts, training data and testing data , and now what is the percentage of the data which are using for raining or testing. In general, there is no specific ratio but the major group of researchers used 75% training to 25% testing. For this ratio, you need to use 75% from your data to training while the rest of data will be using for testing.

Can we able to answer 10 data is enough? We can not able to say yes or no since it is depending upon your problem and number of independent parameters and the results which you get from your ANN.