ClassificationTraining createTrainingConfiguration()
{
ClassificationTraining c = new ClassificationTraining();
c.NbElements = targetTransforms.Length;
c.NbInputs = 2;
c.TrainingExemples = new double[c.NbElements * c.NbInputs];
c.ExpectedResults = new int[c.NbElements];
for (int i = 0; i < targetMeshRenderers.Length; ++i)
{
MeshRenderer mr = targetMeshRenderers[i].GetComponent <MeshRenderer>();
if (mr.material.color == redMat.color)
{
Transform t = targetTransforms[i].GetComponent <Transform>();
c.TrainingExemples[i * c.NbInputs] = t.position.x;
c.TrainingExemples[i * c.NbInputs + 1] = t.position.z;
c.ExpectedResults[i] = 1;
}
if (mr.material.color == blueMat.color)
{
Transform t = targetTransforms[i].GetComponent <Transform>();
c.TrainingExemples[i * c.NbInputs] = t.position.x;
c.TrainingExemples[i * c.NbInputs + 1] = t.position.z;
c.ExpectedResults[i] = -1;
}
}
c.MaxIteration = 100000;
return(c);
}
void ClassificationScenario(ClassificationTraining c, System.IntPtr model, int[] inputneuro)
{
//printModel(model);
trainMLP(c.TrainingExemples, c.NbElements, c.ExpectedResults, model, inputneuro, inputneuro.Length, c.MaxIteration, 0.01);
//printModel(model);
adjustPlan(model, c.NbInputs, inputneuro);
//removeModel(model);
}
