public void ShouldReturnCorrectGenotype()
{
NetVector v1 = NetVector.FromValues(new double[] { 1, 2, 3 });
NetVector v2 = NetVector.FromValues(new double[] { 4, 5, 6 });
NetVector v3 = NetVector.FromValues(new double[] { 7, 8, 9 });
NetVector v4 = NetVector.FromValues(new double[] { 10, 11, 12 });
NetVector v5 = NetVector.FromValues(new double[] { 13, 14, 15, 16, 17 });
NetVector v6 = NetVector.FromValues(new double[] { 18, 19, 20, 21, 22 });
Neuron n1 = new Neuron(v1, Config.ACTIVATION_FUNC);
Neuron n2 = new Neuron(v2, Config.ACTIVATION_FUNC);
Neuron n3 = new Neuron(v3, Config.ACTIVATION_FUNC);
Neuron n4 = new Neuron(v4, Config.ACTIVATION_FUNC);
Neuron n5 = new Neuron(v5, Config.ACTIVATION_FUNC);
Neuron n6 = new Neuron(v6, Config.ACTIVATION_FUNC);
NetLayer l1 = new NetLayer(new Neuron[] { n1, n2, n3, n4 });
NetLayer l2 = new NetLayer(new Neuron[] { n5, n6 });
NeuralNetwork n = new NeuralNetwork(new NetLayer[] { l1, l2 });
List <NeuralNetwork> ln = new List <NeuralNetwork>();
ln.Add(n);
List <List <double> > genotypes = new NeuralNetworkConverter().ToGenotypes(ln);
Assert.AreEqual(genotypes.Count, 1);
TestUtils.AssertEqualDoubleList(
new List <double>(new double[] { 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22 }),
genotypes[0]
);
}
private NeuralNetwork ToNetwork(List <double> genotype, int[] topology)
{
int index = 0;
NetLayer[] layers = new NetLayer[topology.Length - 1];
for (int i = 1; i < topology.Length; ++i)
{
Neuron[] neurons = new Neuron[topology[i]];
for (int j = 0; j < neurons.Length; ++j)
{
double[] weight = new double[topology[i - 1] + 1];
for (int k = 0; k < weight.Length; ++k)
{
weight[k] = genotype[index++];
}
neurons[j] = new Neuron(NetVector.FromValues(weight), Config.ACTIVATION_FUNC);
}
layers[i - 1] = new NetLayer(neurons);
}
return(new NeuralNetwork(layers));
}
public void ShouldReturnCorrectWithoutBias()
{
NetVector v1 = NetVector.FromValues(new double[] { 5, 1.5, 2.4 });
NetVector v2 = NetVector.FromValues(new double[] { 0.23, 3.4, 9.9 });
double result = v1.Scalar(v2);
Assert.AreEqual(30.01, result, 0.01);
}
public void ShouldReturnCorrectWithBias()
{
NetVector v1 = NetVector.FromValues(new double[] { 1, 3, 2 });
NetVector v2 = NetVector.FromValues(new double[] { 1, 4 });
double result = v1.Scalar(v2, -1);
Assert.AreEqual(10, result);
}
public void ShouldReturnCorrectResponse()
{
NetVector v = NetVector.FromValues(new double[] { 1.1, 2.2, 3.3, 4.4, 5.5 });
Neuron n = new Neuron(v, ActivationFunctions.Linear);
NetVector v2 = NetVector.FromValues(new double[] { 2.2, 3.3, 4.4, 5.5 });
double result = n.Response(v2);
Assert.AreEqual(64.24, result, 0.01);
}
public void ShouldReturnException()
{
NetVector v1 = NetVector.FromValues(new double[] { 5, 1.5, 2.4, 6.6 });
NetVector v2 = NetVector.FromValues(new double[] { 0.23, 3.4, 9.9 });
Assert.Throws <System.ArgumentException>(delegate()
{
v1.Scalar(v2);
});
}
public NetVector Response(NetVector inputVector)
{
int layerSize = this.neurons.Length;
double[] outputs = new double[layerSize];
for (int i = 0; i < layerSize; ++i)
{
outputs[i] = neurons[i].Response(inputVector);
}
return(NetVector.FromValues(outputs));
}
public double[] Response(double[] inputValues)
{
if (inputValues.Length != topology[0])
{
throw new System.ArgumentException("Input vector size is not correct");
}
NetVector response = NetVector.FromValues(inputValues).Normalize;
foreach (NetLayer hidden in layers)
{
response = hidden.Response(response);
}
return(response.Values);
}
public void ShouldReturnCorrectResponse()
{
NetVector v1 = NetVector.FromValues(new double[] { 1, 2, 3 });
Neuron n1 = new Neuron(v1, ActivationFunctions.Linear);
NetVector v2 = NetVector.FromValues(new double[] { 4, 5, 6 });
Neuron n2 = new Neuron(v2, ActivationFunctions.Linear);
NetLayer layer = new NetLayer(new Neuron[] { n1, n2 });
NetVector v3 = NetVector.FromValues(new double[] { 7, 8 }); // -1 + 14 + 24 ; -4 + 35 + 48
double[] r = layer.Response(v3).Values;
Assert.AreEqual(2, r.Length);
Assert.AreEqual(37, r[0], 0.01);
Assert.AreEqual(79, r[1], 0.01);
}