void Update()
{
if (regenereaza)
{
n = new neuralnet();
n.genereaza(4, 5, 20, 10, 100);
regenereaza = false;
}
if (mutate)
{
n.mutatie(1);
mutate = false;
}
if (start)
{
l = new List <float>();
l.Add(s.fdreapta);
l.Add(s.fstanga);
l.Add(s.fspate);
l.Add(s.ffata);
List <float> aux = n.proceseaza(l);
m.rl = aux[0];
m.rr = aux[1];
m.mb = aux[2];
m.mf = aux[3];
m.ok = aux[4];
}
}
static Func <double, double> g = (x) => Sin(x); //Function to fit
static void Main()
{
int n = 15; //hidden nodes
var nn = new neuralnet(n);
double x_start = 0;
double x_end = 2 * PI;
int nx = 25;
var xs = new double[nx];
var ys = new double[nx];
WriteLine("Sampled points of test function:");
for (int i = 0; i < nx; i++)
{
xs[i] = (x_end - x_start) / (nx - 1) * i + x_start;
ys[i] = g(xs[i]);
WriteLine($"{xs[i]} \t {ys[i]}");
}
Write("\n\n");
//initial parameters
for (int i = 0; i < n; i++)
{
nn.parameters[3 * i] = (x_end - x_start) / n * i + x_start;
nn.parameters[3 * i + 1] = 1;
nn.parameters[3 * i + 2] = 1;
}
nn.parameters.print("Initial parameters=");
nn.train(xs, ys);
nn.parameters.print("Post training parameters =");
Write("\n\n");
int n_interpol = 100; //number of interpolated points
double x, y, y_der, y_int;
WriteLine("Interpolated points of test function:");
for (int i = 0; i <= n_interpol; i++)
{
x = (x_end - x_start) / n_interpol * i + x_start;
y = nn.feed(x);
y_der = nn.feed_der(x);
y_int = nn.feed_int(x, 0);
WriteLine($"{x} \t {y} \t {y_der} \t {y_int}");
}
}
void newgen()
{
nr++;
for (int k = 0; k < size; k++)
{
for (int l = 0; l < size; l++)
{
if (nval[k] < nval[l])
{
float aux = nval[k];
nval[k] = nval[l];
nval[l] = aux;
neuralnet a = new neuralnet(n[k]);
n[k] = new neuralnet(n[l]);
n[l] = new neuralnet(n[k]);
}
}
}
for (int k = 4; k < size - 2; k++)
{
if (k % 2 == 0)
{
n[k] = new neuralnet(neuralnet.cross(n[parent1], n[parent2], 2));
}
else
{
n[k] = new neuralnet(neuralnet.cross(n[parent2], n[parent1], 2));
}
}
for (int k = size - 2; k < size; i++)
{
n[k].genereaza(4, 5, nsize, npl, w);
}
for (int k = 0; k < size; k++)
{
nval[k] = 0;
}
gen++;
}
public void reset(neuralnet net)
{
start = true;
n = new neuralnet(net);
g.GetComponent <mv>().reset();
}
void Start()
{
n = new neuralnet();
s = g.GetComponent <snezor>();
m = g.GetComponent <mv>();
}
