static void Main(string[] args)
{
double learningRate = 0.1;
double momentum = 0.0;
double sigmoidAlphaValue = 1.0;
// two hidden layers and one output layer
int[] neuronsInFirstLayer = new int[] { 2, 2, 1 };
// prepare learning data
double[][] input = new double[4][];
double[][] output = new double[4][];
input[0] = new double[2];
input[0][0] = 0.0;
input[0][1] = 0.0;
input[1] = new double[2];
input[1][0] = 1.0;
input[1][1] = 0.0;
input[2] = new double[2];
input[2][0] = 0.0;
input[2][1] = 1.0;
input[3] = new double[2];
input[3][0] = 1.0;
input[3][1] = 1.0;
output[0] = new double[1];
output[0][0] = 0.0;
output[1] = new double[1];
output[1][0] = 0.0;
output[2] = new double[1];
output[2][0] = 0.0;
output[3] = new double[1];
output[3][0] = 1.0;
// create multi-layer neural network
ActivationNetwork network = new ActivationNetwork(new BipolarSigmoidFunction(sigmoidAlphaValue), 2, neuronsInFirstLayer);
// create teacher
BackPropagationLearning teacher = new BackPropagationLearning(network);
// set learning rate and momentum
teacher.LearningRate = learningRate;
teacher.Momentum = momentum;
network.Randomize();
// loop
int i = 0;
while (i < 100000)
{
// run epoch of learning procedure
double error = teacher.RunEpoch(input, output) / 4;
i++;
}
Console.WriteLine("Value Input (0, 0): " + network.Compute(new double[2] {
0.0, 0.0
})[0].ToString());
Console.WriteLine("Value Input (0, 1): " + network.Compute(new double[2] {
0.0, 1.0
})[0].ToString());
Console.WriteLine("Value Input (1, 0): " + network.Compute(new double[2] {
1.0, 0.0
})[0].ToString());
Console.WriteLine("Value Input (1, 1): " + network.Compute(new double[2] {
1.0, 1.0
})[0].ToString());
JObject json = new JObject();
json = network.GetJson();
Console.ReadLine();
File.WriteAllText(Environment.GetFolderPath(Environment.SpecialFolder.Desktop) + @"\neuralnet.conf", JsonHelpers.Serialize(json));
}
