public void PredictCross()
{
if (this.model == null)
{
Debug.Log("Create model before");
return;
}
// Call lib to predict test spheres
foreach (var testSphere in testSpheres)
{
var position = testSphere.position;
var transformedPosition = TransformCross(position);
double[] inputs = { transformedPosition.x, transformedPosition.z };
var predicted = LinearClassification.linear_model_predict_classification(this.model.Value, 2, inputs);
position = new Vector3(
position.x,
predicted * (float)0.5,
position.z
);
testSphere.position = position;
}
}
public void PredictXor()
{
if (this.model == null)
{
Debug.Log("Create model before");
return;
}
// Call lib to predict test spheres
foreach (var testSphere in testSpheres)
{
var position = testSphere.position;
double[] inputs = { Math.Pow(position.x + position.z, 2) };
var predicted = LinearClassification.linear_model_predict_classification(this.model.Value, 1, inputs);
position = new Vector3(
position.x,
predicted * (float)0.5,
position.z
);
testSphere.position = position;
}
Debug.Log("Predicted");
}
public void Train()
{
CreateModel(2);
// Call lib to train on the array
var trainingSphereNumber = trainingSpheres.Length;
var trainingInputs = new double[trainingSphereNumber * 2];
var trainingExpectedOutputs = new double[trainingSphereNumber];
for (var i = 0; i < trainingSphereNumber; i++)
{
trainingInputs[i * 2] = trainingSpheres[i].position.x;
trainingInputs[i * 2 + 1] = trainingSpheres[i].position.z;
trainingExpectedOutputs[i] = trainingSpheres[i].position.y;
}
LinearClassification.linear_model_train_classification(this.model.Value, 2, epoch, 0.001, trainingInputs, trainingSphereNumber, trainingExpectedOutputs);
Debug.Log("Model trained !");
}
void CreateModel()
{
this.model = LinearClassification.linear_model_create(2);
Debug.Log("Created model");
}
void CreateModel(int nbDimension)
{
this.model = LinearClassification.linear_model_create(nbDimension);
Debug.Log("Created model");
}
