// Use this for initialization
void Start()
{
int[] size = { 2, 10, 10, 1 };
System.IntPtr coeff = LibWrapper.create_MLP(size, 4);
double[] points = new double[trainningExample.Length * 3];
for (int i = 0; i < trainningExample.Length; i++)
{
points[i * 3] = (trainningExample[i].position.y > 0)?1:-1;
points[i * 3 + 1] = trainningExample[i].position.x / 10;
points[i * 3 + 2] = trainningExample[i].position.z / 10;
}
LibWrapper.mlp_train_classification(coeff, points, trainningExample.Length, trainningExample.Length * 3, 800000);
/*if(ok == 0) {
* Debug.Log("KO");
* } else {
* Debug.Log("OK");
* }*/
foreach (var sphere in sphereTransforms)
{
double[] point = { sphere.position.x / 10, sphere.position.z / 10 };
double result = LibWrapper.mlp_classify(coeff, point, 2);
//Debug.Log(result);
if (result < 0)
{
sphere.position += Vector3.down;
}
else
{
sphere.position += Vector3.up;
}
}
}
public static void Main()
{
try
{
Bool bPointInRect = 0;
Rect myRect = new Rect();
myRect.left = 10;
myRect.right = 100;
myRect.top = 10;
myRect.bottom = 100;
Point myPoint = new Point();
myPoint.x = 50;
myPoint.y = 50;
bPointInRect = LibWrapper.PtInRect(ref myRect, myPoint);
if (bPointInRect == Bool.True)
{
Console.WriteLine("Point lies within the Rect");
}
else
{
Console.WriteLine("Point did not lie within the Rect");
}
}
catch (Exception e)
{
Console.WriteLine("Exception : " + e.Message);
}
}
// Use this for initialization
void Start()
{
int[] size = { 2, 10, 10, 1 };
System.IntPtr coeff = LibWrapper.create_MLP(size, 4);
double[] points = new double[trainningExample.Length * 3];
for (int i = 0; i < trainningExample.Length; i++)
{
points[i * 3] = trainningExample[i].position.y / 10;
points[i * 3 + 1] = trainningExample[i].position.x / 10;
points[i * 3 + 2] = trainningExample[i].position.z / 10;
}
LibWrapper.mlp_train_regression(coeff, points, trainningExample.Length, trainningExample.Length * 3, 800000);
/*if(ok == 0) {
* Debug.Log("KO");
* } else {
* Debug.Log("OK");
* }*/
foreach (var sphere in sphereTransforms)
{
double[] point = { sphere.position.x / 10, sphere.position.z / 10 };
float result = (float)LibWrapper.mlp_regress(coeff, point, 2);
sphere.position += Vector3.up * result * 10;
}
}
// Use this for initialization
void Start()
{
System.IntPtr coeff = LibWrapper.linear_create(3);
double[] points = new double[trainningExample.Length * 3];
for (int i = 0; i < trainningExample.Length; i++)
{
points[i * 3] = trainningExample[i].position.y;
points[i * 3 + 1] = trainningExample[i].position.x;
points[i * 3 + 2] = trainningExample[i].position.z;
}
LibWrapper.linear_train_regression(coeff, points, trainningExample.Length, trainningExample.Length * 3);
foreach (var sphere in sphereTransforms)
{
double[] point = { sphere.position.x, sphere.position.z };
float result = (float)LibWrapper.regress_point(coeff, point, 2);
sphere.position += Vector3.up * result;
}
}
// Use this for initialization
void Start()
{
int nbinputs = 3;
System.IntPtr coeff = LibWrapper.linear_create(nbinputs);
double[] points = new double[trainningExample.Length * 3];
for (int i = 0; i < trainningExample.Length; i++)
{
points[i * 3] = (trainningExample[i].position.y > 0)?1:-1;
points[i * 3 + 1] = trainningExample[i].position.x;
points[i * 3 + 2] = trainningExample[i].position.z;
}
int ok = LibWrapper.linear_train_classification(coeff, points, trainningExample.Length, trainningExample.Length * 3);
if (ok == 0)
{
Debug.Log("KO");
}
else
{
Debug.Log("OK");
}
foreach (var sphere in sphereTransforms)
{
double[] point = { sphere.position.x, sphere.position.z };
double result = LibWrapper.classify_point(coeff, point, 2);
//Debug.Log(result);
if (result < 0)
{
sphere.position += Vector3.down;
}
else
{
sphere.position += Vector3.up;
}
}
}
private void guessWithMLPClassification(System.IntPtr coeff)
{
int nbErrors = 0;
int nbGood = 0;
var maxage = 90.0;
var maxprix = 93.5;
using (var rd = new StreamReader("titanic_test_data.csv"))
{
while (!rd.EndOfStream)
{
var list = new List <double>();
var splits = rd.ReadLine().Split(';');
var vivant = double.Parse(splits[0], System.Globalization.CultureInfo.InvariantCulture);
var classe = double.Parse(splits[1], System.Globalization.CultureInfo.InvariantCulture);
var sex = splits[2];
var age = double.Parse(splits[3], System.Globalization.CultureInfo.InvariantCulture) / maxage;
var prix = double.Parse(splits[4], System.Globalization.CultureInfo.InvariantCulture) / maxprix;
switch ((int)classe)
{
case 1:
list.Add(1);
list.Add(0);
list.Add(0);
break;
case 2:
list.Add(0);
list.Add(1);
list.Add(0);
break;
case 3:
list.Add(0);
list.Add(0);
list.Add(1);
break;
}
if (sex.Equals("male"))
{
list.Add(1);
list.Add(0);
}
else
{
list.Add(0);
list.Add(1);
}
list.Add(age);
list.Add(prix);
double[] inputs = list.ToArray();
double result = LibWrapper.mlp_classify(coeff, inputs, 7);
vivant = (vivant == 0)?-1:1;
result = (result < 0)?-1:1;
if (result == vivant)
{
nbGood++;
}
else
{
nbErrors++;
}
}
double success = ((double)nbGood / (nbGood + nbErrors)) * 100;
Debug.Log("bien trouvé : " + nbGood);
Debug.Log("Erreur : " + nbErrors);
Debug.Log("% réussite : " + success + "%");
}
}
private System.IntPtr trainWithMLPClassification()
{
var list = new List <double>();
var maxage = 90.0;
var maxprix = 93.5;
using (var rd = new StreamReader("titanic_train_data.csv"))
{
while (!rd.EndOfStream)
{
var splits = rd.ReadLine().Split(',');
var vivant = double.Parse(splits[0], System.Globalization.CultureInfo.InvariantCulture);
var classe = double.Parse(splits[1], System.Globalization.CultureInfo.InvariantCulture);
var sex = splits[2];
var age = double.Parse(splits[3], System.Globalization.CultureInfo.InvariantCulture) / maxage;
var prix = double.Parse(splits[4], System.Globalization.CultureInfo.InvariantCulture) / maxprix;
list.Add(vivant == 0?-1:1);
switch ((int)classe)
{
case 1:
list.Add(1);
list.Add(0);
list.Add(0);
break;
case 2:
list.Add(0);
list.Add(1);
list.Add(0);
break;
case 3:
list.Add(0);
list.Add(0);
list.Add(1);
break;
}
if (sex.Equals("male"))
{
list.Add(1);
list.Add(0);
}
else
{
list.Add(0);
list.Add(1);
}
list.Add(age);
list.Add(prix);
}
double[] inputs = list.ToArray();
int[] size = { 8, 50, 50, 1 };
System.IntPtr coeff = LibWrapper.create_MLP(size, 4);
LibWrapper.mlp_train_classification(coeff, inputs, inputs.Length / 8, inputs.Length, 800000);
return(coeff);
}
}
