How to develop a neural network transfer function?

My project is to develop a transfer function that can prescribe a high resolution wind field from a lower resolution field and then train the function to learn micro conditions in a particular region. Then I can use the function to emulate a high resolution field from a coarse one without the need of the real time high resolution. I have very little experience with ANN and I am not sure what to use to develop and then apply such a function? Is it the fitting app? Or should I use the Multilayer Neural Networks and Backpropagation Training? My both high and coarse data are consisting of 3D matrix with exact same size. Each layer of the 3D matrix is taken from a different day. Having a 3D matrix makes thing more confused to me. Sould I put each matrix layer as one vector input?

Answer 1

[Greg Heath]

Use FITNET with vectors obtained from vectorizing layers.

Hope this helps

Thank you for formally accepting my answer

Greg