public INeuralNetwork Create(NeuralNetworkGene genes)
{
var network = _decoratedFactory.Create(genes);
SetUpSynapseMappingForPropogation(network);
return(network);
}
internal INeuralNetwork mate(INeuralNetwork mother, INeuralNetwork father)
{
NeuralNetworkGene motherGenes = mother.GetGenes();
NeuralNetworkGene childFatherGenes = father.GetGenes();
Random random = new Random();
for (int n = 0; n < childFatherGenes.InputGene.Neurons.Count; n++)
{
var neuron = childFatherGenes.InputGene.Neurons[n];
var motherNeuron = motherGenes.InputGene.Neurons[n];
childFatherGenes.InputGene.Neurons[n] = BreedNeuron(neuron, motherNeuron, random);
}
if (childFatherGenes.HiddenGenes.Count >= motherGenes.HiddenGenes.Count)
{
childFatherGenes.HiddenGenes = MateHiddenLayers(childFatherGenes.HiddenGenes, motherGenes.HiddenGenes, random);
}
else
{
childFatherGenes.HiddenGenes = MateHiddenLayers(motherGenes.HiddenGenes, childFatherGenes.HiddenGenes, random);
}
for (int n = 0; n < childFatherGenes.OutputGene.Neurons.Count; n++)
{
var neuron = childFatherGenes.OutputGene.Neurons[n];
var motherNeuron = motherGenes.OutputGene.Neurons[n];
childFatherGenes.OutputGene.Neurons[n] = BreedNeuron(neuron, motherNeuron, random);
}
INeuralNetwork child = _networkFactory.Create(childFatherGenes);
return(child);
}
public INeuralNetwork Mutate(INeuralNetwork network, double mutateChance, out bool didMutate)
{
NeuralNetworkGene childGenes = network.GetGenes();
bool mutated;
didMutate = false;
if (_config.MutateNumberOfHiddenLayers)
{
childGenes = TryAddLayerToNetwork(childGenes, mutateChance, out mutated);
didMutate = mutated;
}
for (int n = 0; n < childGenes.InputGene.Neurons.Count; n++)
{
var neuron = childGenes.InputGene.Neurons[n];
childGenes.InputGene.Neurons[n] = TryMutateNeuron(neuron, mutateChance, out mutated);;
didMutate = didMutate || mutated;
}
for (int h = 0; h < childGenes.HiddenGenes.Count; h++)
{
if (_config.MutateNumberOfHiddenNeuronsInLayer)
{
childGenes.HiddenGenes[h] = TryAddNeuronsToLayer(childGenes, h, mutateChance, out mutated);
didMutate = didMutate || mutated;
}
for (int j = 0; j < childGenes.HiddenGenes[h].Neurons.Count; j++)
{
var neuron = childGenes.HiddenGenes[h].Neurons[j];
childGenes.HiddenGenes[h].Neurons[j] = TryMutateNeuron(neuron, mutateChance, out mutated);
didMutate = didMutate || mutated;
}
}
return(_networkFactory.Create(childGenes));
}
internal LayerGene GetNextLayerGene(NeuralNetworkGene genes, int hiddenLayerIndex)
{
if (hiddenLayerIndex == genes.HiddenGenes.Count - 1)
{
return(genes.OutputGene);
}
else
{
return(genes.HiddenGenes[hiddenLayerIndex + 1]);
}
}
internal LayerGene GetPreviousLayerGene(NeuralNetworkGene genes, int hiddenLayerIndex)
{
if (hiddenLayerIndex == 0)
{
return(genes.InputGene);
}
else
{
return(genes.HiddenGenes[hiddenLayerIndex - 1]);
}
}
internal LayerGene TryAddNeuronsToLayer(NeuralNetworkGene networkGenes, int hiddenLayerIndex, double mutateChance, out bool didMutate)
{
LayerGene hiddenLayer = networkGenes.HiddenGenes[hiddenLayerIndex];
didMutate = false;
while (_random.NextDouble() <= mutateChance)
{
didMutate = true;
//update layer-1 axon terminals
LayerGene previousLayer = GetPreviousLayerGene(networkGenes, hiddenLayerIndex);
foreach (NeuronGene neuron in previousLayer.Neurons)
{
neuron.Axon.Weights.Add(_weightInitializer.InitializeWeight());
}
hiddenLayer.Neurons.Add(GetRandomHiddenNeuronGene(networkGenes, hiddenLayerIndex));
}
return(hiddenLayer);
}
public INeuralNetwork Create(NeuralNetworkGene genes)
{
var mapping = CreateSynapsesFromGenes(genes);
var inputs = GetAllSynapsesFromLayerMapping(mapping[0]);
var outputs = GetAllSynapsesFromLayerMapping(mapping[mapping.Keys.Count - 1]);
ILayer inputLayer = CreateLayerFromGene(genes.InputGene, mapping, 0);
//Hidden layers
IList <ILayer> hiddenLayers = new List <ILayer>();
for (int h = 0; h < genes.HiddenGenes.Count; h++)
{
hiddenLayers.Add(CreateLayerFromGene(genes.HiddenGenes[h], mapping, h + 1));
}
ILayer outputLayer = CreateLayerFromGene(genes.OutputGene, mapping, genes.HiddenGenes.Count + 1);
return(NeuralNetwork.GetInstance(inputs, inputLayer, hiddenLayers, outputLayer, outputs));
}
internal int DetermineNumberOfHiddenNeuronsInLayer(NeuralNetworkGene networkGenes, double mutateChance)
{
int hiddenLayerSize = networkGenes.InputGene.Neurons.Count;
bool increase = !(_random.NextDouble() <= 0.5);
while (_random.NextDouble() <= mutateChance)
{
if (!increase && hiddenLayerSize == 1)
{
break;
}
else if (!increase)
{
hiddenLayerSize--;
}
else
{
hiddenLayerSize++;
}
}
return(hiddenLayerSize);
}
internal NeuralNetworkGene TryAddLayerToNetwork(NeuralNetworkGene genes, double mutateChance, out bool didMutate)
{
NeuralNetworkGene newGenes = genes;
didMutate = false;
while (_random.NextDouble() <= mutateChance)
{
didMutate = true;
int layerToReplace = _random.Next(newGenes.HiddenGenes.Count);
int hiddenLayerSize = DetermineNumberOfHiddenNeuronsInLayer(genes, mutateChance);
//update layer-1 axon terminals
LayerGene previousLayer = GetPreviousLayerGene(newGenes, layerToReplace);
foreach (NeuronGene neuron in previousLayer.Neurons)
{
neuron.Axon.Weights.Clear();
for (int i = 0; i < hiddenLayerSize; i++)
{
neuron.Axon.Weights.Add(_weightInitializer.InitializeWeight());
}
}
LayerGene newLayer = new LayerGene
{
Neurons = new List <NeuronGene>()
};
newGenes.HiddenGenes.Insert(layerToReplace, newLayer);
for (int i = 0; i < hiddenLayerSize; i++)
{
var newNeuron = GetRandomHiddenNeuronGene(newGenes, layerToReplace);
newGenes.HiddenGenes[layerToReplace].Neurons.Add(newNeuron);
}
}
return(newGenes);
}
internal NeuronGene GetRandomHiddenNeuronGene(NeuralNetworkGene networkGenes, int hiddenLayerIndex)
{
var neuronGene = new NeuronGene
{
Axon = new AxonGene
{
Weights = new List <double>(),
ActivationFunction = GetRandomActivationFunction().GetType()
},
Soma = new SomaGene
{
Bias = _weightInitializer.InitializeWeight(),
SummationFunction = GetRandomSummationFunction().GetType()
}
};
//update terminals for current neuron
LayerGene nextlayer = GetNextLayerGene(networkGenes, hiddenLayerIndex);
for (int i = 0; i < nextlayer.Neurons.Count; i++)
{
neuronGene.Axon.Weights.Add(_weightInitializer.InitializeWeight());
}
return(neuronGene);
}
internal LayerGene GetPreviousLayerGene(NeuralNetworkGene genes, int hiddenLayerIndex)
{
if (hiddenLayerIndex == 0)
{
return genes.InputGene;
}
else
{
return genes.HiddenGenes[hiddenLayerIndex - 1];
}
}
internal Dictionary <int, Dictionary <int, IList <Synapse> > > CreateSynapsesFromGenes(NeuralNetworkGene genes)
{
//layer number + position in layer --> list of terminals
var mapping = new Dictionary <int, Dictionary <int, IList <Synapse> > >();
//Input synapses
mapping[0] = CreateSynapseMapLayer(_inputOutputSynapseFactory, genes.InputGene.Neurons.Count, 1);
//Input neuron terminals
mapping[1] = CreateSynapseMapLayerFromLayerGene(_inputOutputSynapseFactory, genes.InputGene);
//Hidden layers
for (int h = 0; h < genes.HiddenGenes.Count; h++)
{
mapping[h + 2] = CreateSynapseMapLayerFromLayerGene(_hiddenSynapseFactory, genes.HiddenGenes[h]);
}
//Output layer
mapping[genes.HiddenGenes.Count + 2] = CreateSynapseMapLayer(_hiddenSynapseFactory, genes.OutputGene.Neurons.Count, 1);
return(mapping);
}
public INeuralNetwork Create(NeuralNetworkGene genes)
{
var mapping = CreateSynapsesFromGenes(genes);
var inputs = GetAllSynapsesFromLayerMapping(mapping[0]);
var outputs = GetAllSynapsesFromLayerMapping(mapping[mapping.Keys.Count - 1]);
ILayer inputLayer = CreateLayerFromGene(genes.InputGene, mapping, 0);
//Hidden layers
IList<ILayer> hiddenLayers = new List<ILayer>();
for (int h = 0; h < genes.HiddenGenes.Count; h++)
{
hiddenLayers.Add(CreateLayerFromGene(genes.HiddenGenes[h], mapping, h + 1));
}
ILayer outputLayer = CreateLayerFromGene(genes.OutputGene, mapping, genes.HiddenGenes.Count + 1);
return NeuralNetwork.GetInstance(inputs, inputLayer, hiddenLayers, outputLayer, outputs);
}
internal Dictionary<int, Dictionary<int, IList<Synapse>>> CreateSynapsesFromGenes(NeuralNetworkGene genes)
{
//layer number + position in layer --> list of terminals
var mapping = new Dictionary<int, Dictionary<int, IList<Synapse>>>();
//Input synapses
mapping[0] = CreateSynapseMapLayer(_inputOutputSynapseFactory, genes.InputGene.Neurons.Count, 1);
//Input neuron terminals
mapping[1] = CreateSynapseMapLayerFromLayerGene(_inputOutputSynapseFactory, genes.InputGene);
//Hidden layers
for (int h = 0; h < genes.HiddenGenes.Count; h++)
{
mapping[h + 2] = CreateSynapseMapLayerFromLayerGene(_hiddenSynapseFactory, genes.HiddenGenes[h]);
}
//Output layer
mapping[genes.HiddenGenes.Count + 2] = CreateSynapseMapLayer(_hiddenSynapseFactory, genes.OutputGene.Neurons.Count, 1);
return mapping;
}
internal NeuralNetworkGene TryAddLayerToNetwork(NeuralNetworkGene genes, double mutateChance, out bool didMutate)
{
NeuralNetworkGene newGenes = genes;
didMutate = false;
while (_random.NextDouble() <= mutateChance)
{
didMutate = true;
int layerToReplace = _random.Next(newGenes.HiddenGenes.Count);
int hiddenLayerSize = DetermineNumberOfHiddenNeuronsInLayer(genes, mutateChance);
//update layer-1 axon terminals
LayerGene previousLayer = GetPreviousLayerGene(newGenes, layerToReplace);
foreach (NeuronGene neuron in previousLayer.Neurons)
{
neuron.Axon.Weights.Clear();
for (int i = 0; i < hiddenLayerSize; i++)
{
neuron.Axon.Weights.Add(_weightInitializer.InitializeWeight());
}
}
LayerGene newLayer = new LayerGene
{
Neurons = new List<NeuronGene>()
};
newGenes.HiddenGenes.Insert(layerToReplace, newLayer);
for (int i = 0; i < hiddenLayerSize; i++)
{
var newNeuron = GetRandomHiddenNeuronGene(newGenes, layerToReplace);
newGenes.HiddenGenes[layerToReplace].Neurons.Add(newNeuron);
}
}
return newGenes;
}
internal NeuronGene GetRandomHiddenNeuronGene(NeuralNetworkGene networkGenes, int hiddenLayerIndex)
{
var neuronGene = new NeuronGene
{
Axon = new AxonGene
{
Weights = new List<double>(),
ActivationFunction = GetRandomActivationFunction().GetType()
},
Soma = new SomaGene
{
Bias = _weightInitializer.InitializeWeight(),
SummationFunction = GetRandomSummationFunction().GetType()
}
};
//update terminals for current neuron
LayerGene nextlayer = GetNextLayerGene(networkGenes, hiddenLayerIndex);
for (int i = 0; i < nextlayer.Neurons.Count; i++)
{
neuronGene.Axon.Weights.Add(_weightInitializer.InitializeWeight());
}
return neuronGene;
}
internal LayerGene GetNextLayerGene(NeuralNetworkGene genes, int hiddenLayerIndex)
{
if (hiddenLayerIndex == genes.HiddenGenes.Count - 1)
{
return genes.OutputGene;
}
else
{
return genes.HiddenGenes[hiddenLayerIndex + 1];
}
}
internal int DetermineNumberOfHiddenNeuronsInLayer(NeuralNetworkGene networkGenes, double mutateChance)
{
int hiddenLayerSize = networkGenes.InputGene.Neurons.Count;
bool increase = !(_random.NextDouble() <= 0.5);
while (_random.NextDouble() <= mutateChance)
{
if (!increase && hiddenLayerSize == 1)
{
break;
}
else if (!increase)
{
hiddenLayerSize--;
}
else
{
hiddenLayerSize++;
}
}
return hiddenLayerSize;
}
internal LayerGene TryAddNeuronsToLayer(NeuralNetworkGene networkGenes, int hiddenLayerIndex, double mutateChance, out bool didMutate)
{
LayerGene hiddenLayer = networkGenes.HiddenGenes[hiddenLayerIndex];
didMutate = false;
while (_random.NextDouble() <= mutateChance)
{
didMutate = true;
//update layer-1 axon terminals
LayerGene previousLayer = GetPreviousLayerGene(networkGenes, hiddenLayerIndex);
foreach (NeuronGene neuron in previousLayer.Neurons)
{
neuron.Axon.Weights.Add(_weightInitializer.InitializeWeight());
}
hiddenLayer.Neurons.Add(GetRandomHiddenNeuronGene(networkGenes, hiddenLayerIndex));
}
return hiddenLayer;
}
