private void autoOptimize(object sender, EventArgs e)
{
var hidden = 3;
var minError = Double.MaxValue;
Parallel.ForEach(new double[] { 0.01, 0.05, 0.1, 0.3, 0.5, 0.7, 0.9, 0.95, 0.96, 0.97, 0.98, 0.99, 1 }, alpha =>
{
foreach (var eta in new double[] { 0.01, 0.05, 0.1, 0.3, 0.5, 0.7, 0.9, 0.95, 0.96, 0.97, 0.98, 0.99, 1 })
{
for (int j = 0; j < 10; j++)
{
var nn = new MLP.MLP(patterns, targets, hidden, alpha, eta);
var error = 0.0;
for (int i = 0; i < 10000; i++)
error = nn.train();
if (error < minError)
{
Console.WriteLine("Alpha =" + alpha + " eta=" + eta + " error=" + error);
minError = error;
}
}
}
});
Console.WriteLine("Done");
}
public SimpleTestDemo()
{
InitializeComponent();
var patterns = new double[1, 1];
var targets = new double[1, 1];
patterns[0, 0] = 0;
targets[0, 0] = 0;
var nn = new MLP.MLP(Matrix<Double>.Build.DenseOfArray(patterns), Matrix<Double>.Build.DenseOfArray(targets), 1, 0, 1.0);
for (int i = 0; i < 2; i++)
{
nn.train();
}
Console.Write("");
}
private void runButton_Click(object sender, EventArgs e)
{
prepareChart();
var patterns = DenseMatrix.OfArray(new double[,] { { 0, 0 }, { 0, 1 }, { 1, 0 }, { 1, 1 }, { 1, 0 }, { 1, 1 } }).Transpose();
var targets = DenseMatrix.OfArray(new double[,] { { 0, 1, 1, 0, 1, 0 } });
var cycles = (int)cyclesNumeric.Value;
var nn = new MLP.MLP(patterns, targets, (int)hiddenNumeric.Value, (double)alphaNumeric.Value, (double)etaNumeric.Value);
for (int i = 0; i < cycles; i++)
{
var error = nn.train();
var result = nn.evaluate(patterns);
for (int j = 0; j < 4; j++)
chart1.Series[j].Points.AddY(result[0, j]);
chart2.Series.Last().Points.AddY(error);
}
}
private void runButton_Click(object sender, EventArgs e)
{
int cycles = prepareCharts();
var nn = new MLP.MLP(patterns, targets, (int)hiddenNumeric.Value, (double)alphaNumeric.Value, (double)etaNumeric.Value);
for (int i = 0; i < cycles; i++)
chart2.Series.Last().Points.AddY(nn.train());
var result = nn.evaluate(patterns);
setLabelText(result);
}