public void Train_RuningTraining_NetworkIsTrained()
{
var network = new SimpleNeuralNetwork(3);
var layerFactory = new NeuralLayerFactory();
network.AddLayer(layerFactory.CreateNeuralLayer(3, new RectifiedActivationFuncion(), new WeightedSumFunction()));
network.AddLayer(layerFactory.CreateNeuralLayer(1, new SigmoidActivationFunction(0.7), new WeightedSumFunction()));
network.PushExpectedValues(
new double[][] {
new double[] { 0 },
new double[] { 1 },
new double[] { 1 },
new double[] { 0 },
new double[] { 1 },
new double[] { 0 },
new double[] { 0 },
});
network.Train(
new double[][] {
new double[] { 150, 2, 0 },
new double[] { 1002, 56, 1 },
new double[] { 1060, 59, 1 },
new double[] { 200, 3, 0 },
new double[] { 300, 3, 1 },
new double[] { 120, 1, 0 },
new double[] { 80, 1, 0 },
}, 10000);
network.PushInputValues(new double[] { 1054, 54, 1 });
var outputs = network.GetOutput();
}
// Update is called once per frame
void FixedUpdate()
{
if (!isDead)
{
if (Input.GetMouseButtonDown(0))
{
Flap();
}
else if (rotation > minAngle && delay == 0)
{
rotation -= rotationSpeed;
}
// Delay rotation of bird after 'Flap'
if (delay > 0)
{
delay--;
}
rb2d.MoveRotation(rotation);
UpdateDistanceMeasure();
//w gore 5.4
//w dol -2.685
//miedzy kolumnami 3.66
double yDistance = (transform.position.y - closestColumn.position.y + 8.085f) / 16.17f;
double xDistance = 0;
if (GameController.instance.score > 0)
{
xDistance = distance / 8.875f;
}
else
{
xDistance = distance / 10.374f;
}
if (xDistance > 1)
{
xDistance = 1;
}
network.PushInputValues(new double[] { (1 - yDistance), xDistance * 0 });
var outputs = network.GetOutput();
//Debug.Log("y: " + yDistance);
//Debug.Log("x: " + xDistance);
//Debug.Log("Output: " + outputs.First());
if (outputs.First() == 1)
{
Flap();
}
}
}
public void PushInputValues_ValuesSentToNetwork_ValuesSetOnInput()
{
var network = new SimpleNeuralNetwork(3);
network.PushInputValues(new double[] { 3, 5, 7 });
Assert.AreEqual(3, network._layers.First().Neurons.First().Inputs.First().GetOutput());
}
public void PushInputValues_ValuesSentToNetwork_ValuesSetOnInput()
{
var network = new SimpleNeuralNetwork(6);
network.PushInputValues(new double[] { 23, 565, 789, 3, 90, 23 });
Assert.AreEqual(23, network._layers.First().Neurons.First().Inputs.First().GetOutput());
Console.WriteLine(JsonConvert.SerializeObject(network, Formatting.Indented));
}
public void AddLayer_NeuralAddingNewLayer_LayerAdded()
{
var network = new SimpleNeuralNetwork(6);
var layerFactory = new NeuralLayerFactory();
network.PushInputValues(new double[] { 23, 565, 789, 3, 90, 23 });
network.AddLayer(layerFactory.CreateNeuralLayer(3, new RectifiedActivationFuncion(), new WeightedSumFunction()));
Assert.AreEqual(2, network._layers.Count);
Console.WriteLine(JsonConvert.SerializeObject(network, Formatting.Indented));
}
static void Main(string[] args)
{
var network = new SimpleNeuralNetwork(1);
var layerFactory = new NeuralLayerFactory();
network.AddLayer(layerFactory.CreateNeuralLayer(2, new RectifiedActivationFuncion(), new WeightedSumFunction()));
network.AddLayer(layerFactory.CreateNeuralLayer(1, new SigmoidActivationFunction(0.4), new WeightedSumFunction()));
double[][] expectedValues = new double[samples][];
double[][] trainingValues = new double[samples][];
for (int i = 0; i < samples; i++)
{
Random rng = new Random();
Random rng2 = new Random();
int val1 = rng.Next(rng.Next() % 1000);
int val2 = rng2.Next(i % 900);
expectedValues[i] = new double[] { (val1 + val2) % 2 };
trainingValues[i] = new double[] { val1, val2 };
Console.WriteLine($"val1: {val1} val2: {val2} sum: { (val1 + val2) % 2 }");
}
network.PushExpectedValues(expectedValues);
network.Train(trainingValues, 5000);
network.PushInputValues(new double[] { 1054, 54 });
var outputs = network.GetOutput();
Console.WriteLine($"network output: {string.Join(", ", outputs)}");
Console.ReadKey();
}
public void TrainNetwork_6Inputs_3HiddenLayer_2Outputs()
{
var network = new SimpleNeuralNetwork(6, 1.95); // six input nuerons
var layerFactory = new NeuralLayerFactory();
network.AddLayer(layerFactory.CreateNeuralLayer(3, new SigmoidActivationFunction(0.7), new WeightedSumFunction())); // three hidden layers
network.AddLayer(layerFactory.CreateNeuralLayer(2, new LazyOutputFunction(), new WeightedSumFunction())); // two output layers
network.PushExpectedValues(
new double[][] {
new double[] { 0.25, 0.20 },
new double[] { 0.10, 0.05 },
new double[] { 0.16, 0.30 },
new double[] { 0.30, 0.10 },
new double[] { 0.25, 0.20 },
new double[] { 0.10, 0.05 },
new double[] { 0.16, 0.30 },
new double[] { 0.30, 0.10 },
});
network.Train(
new double[][] {
new double[] { 150, 0, 0, 34, 35, 56 },
new double[] { 190, 23, 56, 0, 29, 529 },
new double[] { 290, 3, 108, 24, 189, 20 },
new double[] { 290, 67, 6, 0, 1, 0 },
new double[] { 150, 0, 0, 34, 35, 56 },
new double[] { 190, 23, 56, 0, 29, 529 },
new double[] { 290, 3, 108, 24, 189, 20 },
new double[] { 290, 67, 6, 0, 1, 0 },
}, 10000);
network.PushInputValues(new double[] { 150, 0, 0, 34, 35, 56 });
var outputs = network.GetOutput();
Console.WriteLine(outputs[0].ToString() + " " + outputs[1].ToString() + "\n\n" + JsonConvert.SerializeObject(network, Formatting.Indented));
}