I am facing the following problem: Given a list a list of measurements like this random measurements I want to construct an ANN to make up the function. After that I need to be able for a new measurement to get the value of the approximation function. What are the steps I should follow? To my understanding first I need to train the network but how can I achieve that if I don't have any value for output? Second thing how will be I able to reuse the function approximation for a new measurement so it lies on the function approximation curve?
I will use the FANN library.