public void RandomizeWeights()
{
var random = new Random();
RandomizeLayer(InputLayer, (HiddenLayers.FirstOrDefault() ?? OutputLayer).Count(), random);
for (int i = 0; i < HiddenLayers.Count; i++)
{
RandomizeLayer(HiddenLayers[i], (HiddenLayers.ElementAtOrDefault(i + 1) ?? OutputLayer).Count, random);
}
}
public List <double> FeedForward(double[] input)
{
if (input.Length != InputLayer.Count)
{
throw new ArgumentException("input length has to be the same as the InputLayer length");
}
// calculate input activations
for (int i = 0; i < input.Length; i++)
{
InputLayer[i].ActivationValue = input[i];
}
// calculate hidden layer activations
// i : LayerIndex in HiddenLayers
// j : NeuronIndex in HiddenLayers[i]
for (int i = 0; i < HiddenLayers.Count; i++)
{
var previousLayer = HiddenLayers.ElementAtOrDefault(i - 1) ?? InputLayer;
for (int j = 0; j < HiddenLayers[i].Count; j++)
{
HiddenLayers[i][j].ActivationValue = CalculateWeightedSum(previousLayer, j);
}
}
// calculate output activations
// i : NeuronIndex in OutputLayer
var lastLayer = HiddenLayers.LastOrDefault() ?? InputLayer;
for (int i = 0; i < OutputLayer.Count; i++)
{
var currentNeuron = OutputLayer[i];
OutputLayer[i].ActivationValue = CalculateWeightedSum(lastLayer, i);
}
// return only values of output layer
return(OutputLayer.Select(x => x.ActivationValue).ToList());
}
