Any Encog Documentation on NEAT?

Question

Is there any detailed documentation on how to use Encog Framework? I am aware of the user guide available at UserGuide and the C# examples at encog github repository

However I am focused on using NEAT Networks and there is only XOR example available for the same. My so far, code looks like below:

Dim inputNeurons As Integer = 6, outputNeurons As Integer = 2

Dim trainedNetwork As NEATNetwork
Dim network As New NEATPopulation(inputNeurons, outputNeurons, 5000)
network.Reset()
network.InitialConnectionDensity = 100.0   

Dim speciation As New OriginalNEATSpeciation()
speciation.CompatibilityThreshold = 1

Dim trainer As TrainEA
trainer = NEATUtil.ConstructNEATTrainer(network, New networkScore())
trainer.Speciation = speciation

epoch = 1
iterationLimit = 30
i = 1
While i <= iterationLimit
    trainer.Iteration() 
    Console.WriteLine("Iteration #" & i & "/" & iterationLimit & " #" + epoch.ToString + " Score:" + trainer.[Error].ToString)
    epoch += 1
    i += 1
End While

However apart from this there are so many functions available and there is no detailed description on what they do? and how they can be tuned to maximize intelligence for network?

Class NEATPopulation

setInitialConnectionDensity,setInnovations,setRandomNumberFactory,
setSubstrate,setSurvivalRate,setCODEC,
setActivationCycles,setInputCount

and much more in TrainEA Class, NEATNetwork Class etc.

Can anyone share any insight on this network in particular?

Supranational

Answer 1

I organized all of the Encog NEAT documentation into one spot now. You can find it here, with links to several NEAT and HyperNEAT Encog examples. This also describes a fair amount of the NEAT algorithm, from the original papers.

http://www.heatonresearch.com/encog/neat/

Furtive

Further to the above, is there any documentation or examples showing how to incorporate cross validation / k-fold cross validation with Encog NEAT? I am using your XOR example as basis for my project. – Vocalize