/// <summary>
/// Function for predicting the output of the Neural Network, given the inputs, via multiplicating the inputs with the weights of the Network, adding the Bias and then
/// applying an Activation function to the result. This is done until we have reached the last layer, the output layer.
/// </summary>
/// <param name="inputs">The inputs are the speed and distances projected in 5 different dirrections from the car to the next nearest object.</param>
/// <param name="outSteering">The predicted steering.</param>
/// <param name="outAxeleration">The predicted axeleration</param>
public void predict(float[] inputs, ref double outSteering, ref double outAxeleration)
{
double[] firstHiddenLayer = new double[7];
double[] secondHiddenLayer = new double[6];
double[] outputLayer = new double[2];
int i, j;
#region Prediction Algorithm
for (j = 0; j < 7; j++)
{
for (i = 0; i < 6; i++)
{
firstHiddenLayer[j] += inputs[i] * inputWeights[i, j];
}
}
for (i = 0; i < 7; i++)
{
firstHiddenLayer[i] = ActivationFunctions.SigmoidFunction(firstHiddenLayer[i] + inputBias[i]);
}
for (j = 0; j < 6; j++)
{
for (i = 0; i < 7; i++)
{
secondHiddenLayer[j] += firstHiddenLayer[i] * hiddenWeights[i, j];
}
}
for (i = 0; i < 6; i++)
{
secondHiddenLayer[i] = ActivationFunctions.SigmoidFunction(secondHiddenLayer[i] + hiddenBias[i]);
}
for (j = 0; j < 2; j++)
{
for (i = 0; i < 6; i++)
{
outputLayer[j] += secondHiddenLayer[i] * outputWeights[i, j];
}
}
for (i = 0; i < 2; i++)
{
outputLayer[i] = ActivationFunctions.SigmoidFunction(outputLayer[i] + outputBias[i]);
}
outSteering = outputLayer[0];
outAxeleration = outputLayer[1];
#endregion
}
