public void Evolve()
{
List <NeuralNetwork> allNewNetworks = new List <NeuralNetwork>();
List <NeuralNetwork> networkList = networks.Select(kvp => kvp.Value).ToList();
networkList.Sort((t1, t2) => { return(-fitnesses[t1.GetGuid()].CompareTo(fitnesses[t2.GetGuid()])); });
List <NeuralNetwork> best10 = networkList.Take((int)(networkList.Count * 0.1)).ToList();
List <double> fitness = networkList.Select(n => fitnesses[n.GetGuid()]).ToList();
allNewNetworks.AddRange(networkList.Take(5).Select(n => n));
double[] mutationParams = mutationFunction(maxFitness);
allNewNetworks.AddRange(best10.SelectMany(n => n.Mutate(mutationParams[0], mutationParams[1], 4)));
while (allNewNetworks.Count < networkList.Count * 0.75)
{
NeuralNetwork n1 = networkList[RandomValues.RandomInt(0, networkList.Count - 1)];
NeuralNetwork n2 = networkList[RandomValues.RandomInt(0, networkList.Count - 1)];
if (n1.WantsSexyTimeWith(n2) && n2.WantsSexyTimeWith(n1))
{
allNewNetworks.Add(n1.DoSexyTimeWith(n2));
}
}
while (allNewNetworks.Count < networkList.Count)
{
int[] layers = new int[RandomValues.RandomInt(layerLength.Item1, layerLength.Item2)];
for (int i = 0; i < layers.Length; i++)
{
layers[i] = RandomValues.RandomInt(layerRange.Item1, layerRange.Item2);
}
layers[0] = inputs;
layers[layers.Length - 1] = outputs;
NeuralNetwork network = new NeuralNetwork(layers, activationFunction);
allNewNetworks.Add(network);
}
networks.Clear();
fitnesses.Clear();
foreach (NeuralNetwork network in allNewNetworks)
{
networks.Add(network.GetGuid(), network);
}
generation++;
}
public void Init()
{
for (int pop = 0; pop < netPopulations; pop++)
{
int[] layers = new int[RandomValues.RandomInt(layerLength.Item1, layerLength.Item2)];
for (int i = 0; i < layers.Length; i++)
{
layers[i] = RandomValues.RandomInt(layerRange.Item1, layerRange.Item2);
}
layers[0] = inputs;
layers[layers.Length - 1] = outputs;
for (int net = 0; net < netsPerPopulation; net++)
{
NeuralNetwork network = new NeuralNetwork(layers, activationFunction);
networks.Add(network.Guid, network);
}
}
}