public void PropagateSignal()
{
var levelsRight = new Dictionary<double, IntervalD>
{
{0.0, new IntervalD(1.0, 3.0)},
{0.5, new IntervalD(1.5, 2.5)},
{1.0, new IntervalD(2.0, 2.0)},
};
var weight = new DiscreteFuzzyNumber(levelsRight);
var input = new Link(weight);
var output = new Link();
var f = new Func<double, double>(x => 1.0 / (1.0 + Math.Pow(Math.E, -x)));
var neuron = new FuzzyNeuron(f, () => DiscreteFuzzyNumber.GenerateLittleNumber(levelsCount: 3));
neuron.AddInput(input, weight);
neuron.SetOutput(output);
neuron.Propagate();
var expectedLevel00 = new IntervalD(f(1), f(9));
var expectedLevel05 = new IntervalD(f(2.25), f(6.25));
var expectedLevel10 = new IntervalD(f(4.0), f(4.0));
Assert.NotNull(output.Signal);
Assert.That(output.Signal.GetAlphaLevel(0.0), Is.EqualTo(expectedLevel00));
Assert.That(output.Signal.GetAlphaLevel(0.5), Is.EqualTo(expectedLevel05));
Assert.That(output.Signal.GetAlphaLevel(1.0), Is.EqualTo(expectedLevel10));
}
public SimpleFuzzyNet(int inputNeuronsCount, int[] hiddenNeuronsCount, Func<IFuzzyNumber> littleFuzzyNumberGenerator, Func<double, double> activationFunction = null, int outputNeuronsCount = 1, int levelsCount = 100)
{
for (int i = 0; i < inputNeuronsCount; i++)
{
var inputSignal = new Link();
_inputSignals.Add(inputSignal);
}
for (int i = 0; i < hiddenNeuronsCount.Length; i++)
{
_hiddens.Add(new Layer(hiddenNeuronsCount[i],
levelsCount,
littleFuzzyNumberGenerator,
activationFunction) {Name = "Hidden layer " + i});
}
_output = new Layer(outputNeuronsCount, levelsCount, littleFuzzyNumberGenerator, activationFunction){Name = "Output layer"};
Layers = new List<ILayer>(_hiddens) {_output};
for (int i = 0; i < outputNeuronsCount; i++)
{
var outputSignal = new Link();
_outputSignals.Add(outputSignal);
}
BuildStructure();
}
public void ConnectTo(ILayer nextLayer)
{
_forwardLinks.Clear();
foreach (var neuron in _neurons)
{
var output = new Link();
_forwardLinks.Add(output);
neuron.SetOutput(output);
nextLayer.ForeachNeuron((i, nextNeuron) => nextNeuron.AddInput(output));
}
}
