public override WeightsMatrix WeightsUpdate(WeightsMatrix gradient)
{
WeightsMatrix result = gradient.Copy();
result.Scale(-_stepSize / _batchSize);
return(result);
}
public bool Equals(WeightsMatrix other)
{
if (other == null)
{
return(false);
}
if (this.NumberOfInputs != other.NumberOfInputs || this.NumberOfOutputs != other.NumberOfOutputs)
{
return(false);
}
double epsilon = 0.000000001;
for (int i = 0; i < this.NumberOfOutputs; i++)
{
for (int j = 0; j < NumberOfInputs; j++)
{
double difference = Math.Abs(this._matrix[i, j] - other._matrix[i, j]);
if (difference >= epsilon)
{
return(false);
}
}
}
return(true);
}
public Layer(
WeightsMatrix weights,
NetworkVector biases,
ActivationFunction activationfunction,
DerivativeFunction derivativefunction
)
: base(weights.NumberOfOutputs, weights.NumberOfInputs)
{
if (activationfunction != null && derivativefunction == null)
{
throw new ArgumentException("derivativefunction cannot be null, if activatioin is not null");
}
if (weights == null || biases == null)
{
throw new ArgumentException("Attempt to make a layer with null weights or biases.");
}
_combiner = new WeightedCombiner(weights, biases);
if (activationfunction == null)
{
_neuralFunction = null;
}
else
{
_neuralFunction = new NeuralFunction(_combiner.NumberOfOutputs, activationfunction, derivativefunction);
}
}
public override WeightsMatrix WeightsGradient(NetworkVector outputgradient)
{
WeightsMatrix weightsGradient = new WeightsMatrix(Weights.NumberOfOutputs, Weights.NumberOfInputs);
foreach (var pair in _segmentAndPair(VectorInput, outputgradient))
{
weightsGradient.Add(pair.Second.OuterProduct(pair.First));
}
return(weightsGradient);
}
public WeightedCombiner(WeightsMatrix weights, NetworkVector biases)
: base(weights.NumberOfOutputs, weights.NumberOfInputs)
{
if (weights == null)
{
throw new ArgumentException("Attempt to make a WeightedCombineer with weights == null.");
}
if (biases == null)
{
throw new ArgumentException("Attempt to make a WeightedCombineer with biases == null.");
}
if (biases.Dimension != weights.NumberOfOutputs)
{
throw new ArgumentException("Dimension of biases must the the same of the outputs.");
}
Weights = weights.Copy();
Biases = biases.Copy();
VectorInput = new NetworkVector(weights.NumberOfInputs);
//Output = new NetworkVector(weights.NumberOfOutputs);
}
public static Layer CreateLinearLayer(WeightsMatrix weights, NetworkVector biases)
{
return(new Layer(weights, biases, null, null));
}
public static Layer CreateLogisticLayer(WeightsMatrix weights)
{
return(new Layer(weights, new NetworkVector(weights.NumberOfOutputs), NeuralFunction.__Logistic, NeuralFunction.__LogisticDerivative));
}
public Layer(WeightsMatrix weights)
: this(weights, new NetworkVector(weights.NumberOfOutputs), null, null)
{
}
public Layer(WeightsMatrix weights, NetworkVector biases)
: this(weights, biases, null, null)
{
}
public static Layer CreateLogisticLayer(WeightsMatrix weights, NetworkVector biases)
{
return(new Layer(weights, biases, NeuralFunction.__Logistic, NeuralFunction.__LogisticDerivative));
}
public abstract WeightsMatrix WeightsUpdate(WeightsMatrix gradient);
public static Layer CreateLinearLayer(WeightsMatrix weights)
{
return(new Layer(weights, new NetworkVector(weights.NumberOfOutputs), null, null));
}
public WeightedCombiner(WeightedCombiner combiner)
: base(combiner.NumberOfOutputs, combiner.NumberOfInputs)
{
this.Biases = combiner.Biases.Copy();
this.Weights = combiner.Weights.Copy();
}
public WeightedCombiner(WeightsMatrix weights)
: this(weights, new NetworkVector(weights.NumberOfOutputs))
{
}
public void Add(WeightsMatrix other)
{
_matrix = _matrix.Add(other._matrix);
}
public void Subtract(WeightsMatrix other)
{
_matrix = _matrix.Subtract(other._matrix);
}
public TrainableComponent(int numberofoutputs, int numberofinputs)
{
_weightsGradientAccumulator = new WeightsMatrix(numberofoutputs, numberofinputs);
_biasesGradientAccumulator = new NetworkVector(numberofoutputs);
}
