//static public Brain GetBrain(Creature owner, Creature parent)
//{
// Brain b = new Brain(owner);
// #region set brain size
// int HLayers = owner.stats.hiddenLayers;
// b.SetHiddenLayers(HLayers);
// #endregion
// #region set gene order
// HashSet<string> pheroHash = new HashSet<string>();
// var geneOrder = new string[pheroHash.Count];
// int index = 0;
// foreach (var s in pheroHash)
// {
// geneOrder[index] = s;
// index++;
// }
// b.SetGeneOrder(geneOrder);
// #endregion
// #region create layers
// BrainLayer inputBL;
// int VSize = owner.stats.vEye;
// //var HGenes = owner.GetGenes(Creature.GeneTypes.hearing);
// int HSize = owner.stats.eEars;
// var PGenes = owner.GetGenes(Creature.GeneTypes.pheromone);
// foreach (var g in PGenes)
// {
// pheroHash.Add(g.GetGeneticString().Substring(3));
// }
// int outputSize = 7;
// int inputSize = (HSize * 3) + (VSize * 3) + (PGenes.Count * 2);
// //inputBL = new BrainLayer(parent.brain.inputs, inputSize, owner.stats.hiddenSize, owner.stats.tweakPercentage,7+ (HSize * 3) + (VSize * 3), geneOrder,parent.brain.PheromoneOrder);
// BrainLayer[] hiddenBL=new BrainLayer[HLayers];
// if (parent.brain.hiddenLayerCount < HLayers)
// {
// for (int i = 0; i < parent.brain.hiddenLayerCount - 1; i++)
// {
// hiddenBL[i] = new BrainLayer(parent.brain.hiddens[i], owner.stats.hiddenSize, owner.stats.hiddenSize, owner.stats.tweakPercentage);
// }
// //fill missing -1
// for (int i= parent.brain.hiddenLayerCount;i< HLayers - 1; i++)
// {
// hiddenBL[i] = new BrainLayer(owner.stats.hiddenSize, owner.stats.hiddenSize);
// }
// }
// else
// {
// for (int i = 0; i < HLayers-1; i++)
// {
// hiddenBL[i] = new BrainLayer(parent.brain.hiddens[i],owner.stats.hiddenSize, owner.stats.hiddenSize, owner.stats.tweakPercentage);
// }
// }
// /////hiddentBL[last]= new BL(parent.brain.hiddens[last]
// //int outputSize = 7;
// //outputSize += owner.GetGenes(Creature.GeneTypes.outputGeneP).Count;
// //hiddenBL[HLayers - 1] = new BrainLayer(owner.stats.hiddenSize, outputSize);
// BrainLayer outputBL;
// outputSize += owner.GetGenes(Creature.GeneTypes.outputGeneP).Count;
// outputBL = new BrainLayer(parent.brain.outputs, outputSize, 1, owner.stats.tweakPercentage);
// //b.SetLayers(inputBL,hiddenBL,outputBL);
// #endregion
// #region set input start index
// b.SetInputLength(VSize, HSize);
// #endregion
// return b;
//}
//static public Brain GetBrain(Creature owner)
//{
// Brain b = new Brain(owner);
// #region set brain size
// int HLayers = owner.stats.hiddenLayers;
// b.SetHiddenLayers(HLayers);
// #endregion
// #region create layers
// int inputSize = 2;
// //var VGenes = owner.GetGenes(Creature.GeneTypes.vision);
// int VSize = owner.stats.vEye;
// //var HGenes = owner.GetGenes(Creature.GeneTypes.hearing);
// int HSize = owner.stats.eEars;
// var PGenes = owner.GetGenes(Creature.GeneTypes.pheromone);
// HashSet<string> pheroHash = new HashSet<string>();
// foreach (var g in PGenes)
// {
// pheroHash.Add(g.GetGeneticString().Substring(3));
// }
// inputSize += (HSize * 3) + (VSize * 3) + (PGenes.Count * 2);
// var hiddens = new BrainLayer[HLayers];
// for (int i = 0; i < HLayers-1; i++)
// {
// hiddens[i] = new BrainLayer(owner.stats.hiddenSize, owner.stats.hiddenSize);
// }
// int outputSize = 7;
// outputSize += owner.GetGenes(Creature.GeneTypes.outputGeneP).Count;
// hiddens[HLayers - 1] = new BrainLayer(owner.stats.hiddenSize, outputSize);
// b.SetLayers(new BrainLayer(inputSize,owner.stats.hiddenSize),hiddens,new BrainLayer(outputSize,1));
// #endregion
// #region set gene order
// var geneOrder = new string[pheroHash.Count];
// int index = 0;
// foreach (var s in pheroHash)
// {
// geneOrder[index] = s;
// index++;
// }
// b.SetGeneOrder(geneOrder);
// #endregion
// #region set input start index
// b.SetInputLength(VSize, HSize);
// #endregion
// return b;
//}
static public Brain OneSizeBrain(Creature owner)
{
Brain b = new Brain(owner);
#region set brain size
int HLayers = OSBHLayersNumber;
b.SetHiddenLayers(HLayers);
#endregion
#region create layers
int inputSize = 2;
//var VGenes = owner.GetGenes(Creature.GeneTypes.vision);
int VSize = OSBEyesNumber;
//var HGenes = owner.GetGenes(Creature.GeneTypes.hearing);
int HSize = OSBEarsNumber;
inputSize += (HSize * 3) + (VSize * 3);
var hiddens = new BrainLayer[HLayers];
for (int i = 0; i < HLayers - 1; i++)
{
hiddens[i] = new BrainLayer(OSBHSize, OSBHSize);
}
int outputSize = 7;
hiddens[HLayers - 1] = new BrainLayer(OSBHSize, outputSize);
b.SetLayers(new BrainLayer(inputSize, OSBHSize), hiddens, new BrainLayer(outputSize, 1));
#endregion
b.FillBrain(BrainLayer.NNFillingMode.random);
return(b);
}
static public Brain OneSizeBrain(Creature owner, Creature parent)
{
Brain b = new Brain();
#region set brain size
int HLayers = OSBHLayersNumber;
b.SetHiddenLayers(HLayers);
#endregion
#region create layers
BrainLayer inputBL;
inputBL = new BrainLayer(parent.brain.inputs, owner.stats.tweakPercentage);
BrainLayer[] hiddenBL = new BrainLayer[HLayers];
for (int i = 0; i < HLayers; i++)
{
hiddenBL[i] = new BrainLayer(parent.brain.hiddens[i], owner.stats.tweakPercentage);
}
BrainLayer outputBL;
outputBL = new BrainLayer(parent.brain.outputs, owner.stats.tweakPercentage);
b.SetLayers(inputBL, hiddenBL, outputBL);
#endregion
return(b);
}
public BrainLayer(BrainLayer parent, int size, int nextSize, float tweakPercentage)
{
nodes = new float[size, 1];
weights = new float[size, nextSize];
biases = new float[size, 1];
weights = DuplicateArray(parent.weights, size, nextSize, tweakPercentage);
biases = DuplicateArray(parent.biases, size, 1, tweakPercentage);
//bool xPSmaller = false;
//bool yPSmaller = false;
//if(parent.weights.GetLength(0)< weights.GetLength(0))
//{
// xPSmaller = true;
//}
//if(parent.weights.GetLength(1)<weights.GetLength(0))
//{
// yPSmaller = true;
//}
//weights = DuplicateArray(parent.weights,tweakPercentage);
//biases = DuplicateArray(parent.biases, tweakPercentage);
}
public BrainLayer(BrainLayer parent, int size, int nextSize, float tweakPercentage, int parralleStart, int parentStart, string[] ownerOrder, string[] parentOrder)
{
nodes = new float[size, 1];
weights = new float[size, nextSize];
biases = new float[size, 1];
weights = DuplicateArray(parent.weights, size, nextSize, tweakPercentage);
biases = DuplicateArray(parent.biases, size, 1, tweakPercentage);
//foreach order in child
for (int i = 0; i < ownerOrder.Length; i++)
{
//find corresponding
for (int j = 0; j < parentOrder.Length; j++)
{
//if found
if (ownerOrder[i] == parentOrder[j])
{
parentOrder[j] = "";
if (!(i == j))
{
//switch row
var temp = new string[nextSize];
for (int k = 0; k < temp.Length && k < parent.biases.GetLength(0); k++)
{
Swap(ref weights[i + parralleStart, k], ref parent.weights[i + parentStart, k]);
}
Swap(ref biases[i + parralleStart, 0], ref parent.weights[i + parentStart, 0]);
}
}
}
}
//weights = DuplicateArray(parent.weights, tweakPercentage);
//biases = DuplicateArray(parent.biases, tweakPercentage);
}
public void SetLayers(BrainLayer input, BrainLayer[] hidden, BrainLayer output)
{
inputs = input;
hiddens = hidden;
outputs = output;
}
public void FeedForward(BrainLayer previous)
{
nodes = Sigmoid(Add(DotProduct(Transpose(previous.weights), previous.nodes), previous.biases));
}
public BrainLayer(BrainLayer parent, float tweakPercentage)
{
nodes = new float[parent.nodes.GetLength(0), 1];
weights = DuplicateArray(parent.weights, tweakPercentage);
biases = DuplicateArray(parent.biases, tweakPercentage);
}
