public void TestBias()
{
double input = 0.3;
double inpToOutWeight = 0.4;
double biasToOutWeight = 0.5;
var sigmoid = new SigmoidFunction();
var nw = new Network(sigmoid, false);
var inputNode = new Input(input);
var bias = new Bias();
var outputNode = new Perceptron(sigmoid, "Output");
Connection.Create(inpToOutWeight, inputNode, outputNode);
Connection.Create(biasToOutWeight, bias, outputNode);
nw.Nodes = new Node[][] {
new Node[] { inputNode },
new Node[] { outputNode }
};
var nwOut = nw.Nodes[1][0].Output;
//Output
// = Sig(inpToOut.Output + biasToOut.Output)
// = Sig((inpToOut.Weight * inpToOut.GetInput()) + (biasToOutWeight * 1))
// = Sig((inpToOut.Weight * inputNode.Output) + (biasToOutWeight))
// = Sig((this.weight * this.input) + (biasToOutWeight))
var expOut = sigmoid.Calculate((inpToOutWeight * input) + biasToOutWeight);
Assert.AreEqual(nwOut, expOut);
}
public void TestConnection()
{
double weight = 2;
double input = 5;
var inp = new Input(input);
var outp = new Perceptron(new SigmoidFunction(), "Output");
var wc = Connection.Create(weight, inp, outp);
Assert.AreEqual(wc.Output, weight * input);
Assert.AreEqual(inp.GetOutgoingConnections()[0], outp.GetIncommingConnections()[0]);
}
public void TestDelConnection()
{
var inp = new Input(5);
var outp = new Perceptron(new SigmoidFunction());
var beforeConnect = outp.Output;
var con = Connection.Create(1, inp, outp);
outp.ResetCache();
var afterConnect = outp.Output;
Assert.AreNotEqual(beforeConnect, afterConnect);
con.Delete();
outp.ResetCache();
Assert.AreEqual(beforeConnect, outp.Output);
}
public void FillNetwork(int nrInputs, int nrOutputs, params int[] hiddenLayerHeights)
{
//Creation of outputs is similar to creating hidden layers
var nonInputLayers = hiddenLayerHeights.ToList();
nonInputLayers.Add(nrOutputs);
//Init array
Nodes = new Node[nonInputLayers.Count + 1][];
//Inputs
Nodes[0] = CreateInputs(nrInputs).ToArray();
//Hidden & out
for(int layerIndex = 1; layerIndex <= nonInputLayers.Count; layerIndex++) {
int height = nonInputLayers[layerIndex - 1];
var curLayer = new List<Perceptron>(height);
for(int percNr = 0; percNr < height; percNr++) {
//Create perceptron
string name = layerIndex == Nodes.Length - 1 ? $"Output {percNr}" : $"Hidden #{layerIndex - 1}.{percNr}";
var newPerceptron = new Perceptron(TransferFunction, name);
//Create input connections
foreach(var inp in Nodes[layerIndex - 1]) {
double weight = 0;
if(layerIndex - 1 == 0) { //Input --> 1st hidden
weight = MathHelper.GuassianRandom(Math.Sqrt(1.0 / 3.0), 0);
}
Connection.Create(weight, inp, newPerceptron);
}
if(Bias != null) {
Connection.Create(0.5, Bias, newPerceptron);
}
curLayer.Add(newPerceptron);
}
Nodes[layerIndex] = curLayer.ToArray();
}
}
public void TestPerceptronCaching()
{
double local1 = -10;
double local2 = -5;
var in1 = new Input(local1, "Input1");
var in2 = new Input(local2, "Input2");
var p = new Perceptron(new SigmoidFunction());
Connection.Create(0.5, in1, p);
Connection.Create(1, in2, p);
Assert.IsTrue(p.Output < 0.1);
in1.Value = 10;
in2.Value = 5;
Assert.IsTrue(p.Output < 0.1);
p.ResetCache();
Assert.IsFalse(p.Output < 0.1);
}
public void TestNetworkOutput()
{
var rand = new Random();
double input = (double)(rand.NextDouble());
double weight = (double)(rand.NextDouble());
var sigmoid = new SigmoidFunction();
var nw = new Network(sigmoid, false);
var inputNode = new Input(input);
var outputNode = new Perceptron(sigmoid, "Output");
Connection.Create(weight, inputNode, outputNode);
nw.Nodes = new Node[][] {
new Node[] { inputNode },
new Node[] { outputNode }
};
var nwOut = nw.CurOutput()[0];
Assert.AreEqual(outputNode.Output, nwOut);
var expOut = sigmoid.Calculate(input * weight);
Assert.AreEqual(nwOut, expOut);
}
public void TestGetInputResult()
{
var rand = new Random();
double input = (double)(rand.NextDouble() * 100 + 1);
double weight = (double)(rand.NextDouble() * 2);
var sigmoid = new SigmoidFunction();
var nw = new Network(sigmoid, false);
var inputNode = new Input(input);
var outputNode = new Perceptron(sigmoid, "Output");
Connection.Create(weight, inputNode, outputNode);
nw.Nodes = new Node[][] {
new Node[] { inputNode },
new Node[] { outputNode }
};
var nwOut = nw.GetInputResult(input)[0];
//Output
// = Sig(inpToOut.Output)
// = Sig(inpToOut.Weight * inpToOut.GetInput())
// = Sig(inpToOut.Weight * inputNode.Output)
// = Sig(this.weight * this.input)
var expOut = sigmoid.Calculate(weight * input);
Assert.AreEqual(nwOut, expOut);
}
