private static void RecurseNegativeSample(Layer layer, Layer previousLayer, int inputIndex, Dictionary <Node, double> backwardsPassDeltas, double momentumMagnitude)
{
if (!previousLayer.PreviousLayers.Any())
{
// case where previous layer is input
NegativeSampleFirstHiddenLayer(layer, previousLayer, inputIndex, backwardsPassDeltas, momentumMagnitude);
return;
}
var deltas = new Dictionary <Node, double>();
foreach (var node in layer.Nodes)
{
var delta = Backpropagation.CalculateDelta(layer, backwardsPassDeltas, node);
deltas.Add(node, delta);
foreach (var(prevNode, weightForPrevNode) in node.Weights)
{
Backpropagation.UpdateNodeWeight(prevNode, weightForPrevNode, delta, momentumMagnitude);
}
foreach (var(_, weightForPrevLayer) in node.BiasWeights)
{
Backpropagation.UpdateBiasNodeWeight(weightForPrevLayer, delta, momentumMagnitude);
}
}
foreach (var prevPrevLayer in previousLayer.PreviousLayers)
{
RecurseNegativeSample(previousLayer, prevPrevLayer, inputIndex, deltas, momentumMagnitude);
}
}
private static void NegativeSampleFirstHiddenLayer(Layer layer, Layer inputLayer, int inputIndex, Dictionary <Node, double> backwardsPassDeltas, double momentumMagnitude)
{
var inputNode = inputLayer.Nodes[inputIndex];
foreach (var node in layer.Nodes)
{
var delta = Backpropagation.CalculateDelta(layer, backwardsPassDeltas, node);
Backpropagation.UpdateNodeWeight(inputNode, node.Weights[inputNode], delta, momentumMagnitude);
if (node.BiasWeights.TryGetValue(inputLayer, out var biasWeight))
{
Backpropagation.UpdateBiasNodeWeight(biasWeight, delta, momentumMagnitude);
}
}
}
private static Dictionary <Node, double> NegativeSampleOutput(Layer outputLayer, double targetOutput, int outputIndex, ErrorFunctionType errorFunctionType, double learningRate, double momentumMagnitude)
{
var errorFunctionDifferential = ErrorFunctionResolver.ResolveErrorFunctionDifferential(errorFunctionType);
var outputNode = outputLayer.Nodes[outputIndex];
var delta = errorFunctionDifferential.Invoke(targetOutput, outputNode.Output)
* outputLayer.ActivationFunctionDifferential(outputNode.Output)
* learningRate;
foreach (var(prevNode, weightForPrevNode) in outputNode.Weights)
{
Backpropagation.UpdateNodeWeight(prevNode, weightForPrevNode, delta, momentumMagnitude);
}
foreach (var(_, weightForPrevLayer) in outputNode.BiasWeights)
{
Backpropagation.UpdateBiasNodeWeight(weightForPrevLayer, delta, momentumMagnitude);
}
return(new Dictionary <Node, double> {
{ outputNode, delta }
});
}