How to build a not fully-connected neural network step-by-step?

Hi,

are there any methods to develop your network step-by-step?

Something like the Cascade-Correlation-Algorithm where you insert a single neuron after another while you had trained the first one?

I just have found the function for building a casdadenet or a normal feedforwardnet. The problem is, that in these networks all neurons are included.

I want to build the network from the ground up independent and as I want it.

Any thoughts? :)

Thank you.

Answer 1

[BERGHOUT Tarek]

you mean like this one, your question is big , i did it once; but i can give you refrences : read about ELM-LRF