/// <summary>
/// Finds the value which maximizes
/// the activation of this neuron.
/// </summary>
///
public double[] Maximize()
{
// Initialize
double[] value = new double[neuron.InputsCount];
for (int i = 0; i < value.Length; i++)
{
value[i] = Accord.Math.Tools.Random.NextDouble();
}
double[] gradient = new double[neuron.InputsCount];
for (int iter = 0; iter < 50; iter++)
{
// Compute the activation gradient
double activation = neuron.Compute(value);
for (int i = 0; i < gradient.Length; i++)
{
gradient[i] = neuron.Weights[i] * neuron.ActivationFunction.Derivative2(activation);
}
// Walk against the gradient
for (int i = 0; i < value.Length; i++)
{
value[i] -= gradient[i];
}
value = value.Divide(value.Sum());
}
return(value);
}
public void NeuronUpdateWeights()
{
double learningRate = 0.3;
double delta = -0.3;
int InputCount = 3;
double outputValue = 0;
double outputAfterUpdate = 0;
IActivationFunction ActivationSigmoid = new SigmoidFunction();
Neuron ActNeuron = new ActivationNeuron(InputCount, ActivationSigmoid);
ActNeuron.FeedForward(InputValues);
outputValue = ActNeuron.Compute();
Assert.True(outputValue != 0);
ActNeuron.UpdateWeight(learningRate, delta);
outputAfterUpdate = ActNeuron.Compute();
Assert.NotEqual(outputValue, outputAfterUpdate);
}
public void NeuronInitalizeActivationSigmoid()
{
int InputCount = 3;
double outputValue = 0;
IActivationFunction ActivationSigmoid = new SigmoidFunction();
Neuron ActNeuron = new ActivationNeuron(InputCount, ActivationSigmoid);
ActNeuron.FeedForward(InputValues);
outputValue = ActNeuron.Compute();
Assert.True(outputValue != 0);
}
