private void button2_Click(object sender, EventArgs e)
{
Random rng = new Random();
int iter = (int)(numIterations.Value);
List <NormalizedDonor> norms = new List <NormalizedDonor>();
for (int i = 0; i < iter; i++)
{
NormalizedDonor d = TrainingDonors[rng.Next(TrainingDonors.Count)];
norms.Add(d);
double[] ins = { d.Year, d.OperationYear, d.AxillaryNodes };
double[] ots = { d.Status };
nn.Train(new List <double>(ins), new List <double>(ots));
}
int hits = 0;
foreach (NormalizedDonor n in norms)
{
double[] ins = { n.Year, n.OperationYear, n.AxillaryNodes };
double rez = nn.Run(new List <double>(ins))[0];
if (Math.Abs(rez - n.Status) < 0.5)
{
hits++;
}
}
lblTrainAcc.Text = String.Format("Paskutinio apmokymo tikslumas: {0:0.00}%", (100.0 * hits) / iter);
NetworkHelper.ToTreeView(treeView1, nn);
NetworkHelper.ToPictureBox(pictureBox1, nn, 400, 100);
}
private void button2_Click(object sender, EventArgs e)
{
List <double> ins = new List <double>();
ins.Add(0);
ins.Add(1);
List <double> ots = new List <double>();
ots.Add(0);
//ots.Add(0);
for (int i = 0; i < 100000; i++)
{
nn.Train(ins, ots);
}
NetworkHelper.ToTreeView(treeView1, nn);
NetworkHelper.ToPictureBox(pictureBox1, nn, 400, 100);
}
static void Main(string[] args)
{
string result_file = @"\\Mac\Home\Desktop\mnist\result.csv";
List <double> accuracyValues = new List <double>();
double[] learningRates = { 0.2, 0.01, 0.1, 0.02, 0.3, 0.03, 0.4, 0.04 };
double learningRate = 0.1;
// NeuralNetwork neuralnet = new NeuralNetwork(784, 10, 1, 45);
Tuple <double, double> loss_accuracy;
List <Tuple <int, double, Tuple <double, double>, Tuple <double, double> > > result = new List <Tuple <int, double, Tuple <double, double>, Tuple <double, double> > >();
//double accuracy = 0;
string format = "{0, -15}{1,-15}{2,-15}{3, -25}{4, -25}{5, -25}";
string[] heading = new string[] { "Epoch", "Learning Rate", "Training Loss", "Testing Loss", "Training Accuracy", "Testing Accuracy" };
Console.WriteLine(string.Format(format, heading));
foreach (var learning_rate in learningRates)
{
NeuralNetwork neuralnet = new NeuralNetwork(784, 10, 1, 45);
for (var epoch = 0; epoch < 10; ++epoch)
{
neuralnet.Train(MnistReader.TrainImagesPath, MnistReader.TrainLabelsPath, learning_rate);
// evaluate on training data
var trainingMetric = neuralnet.evaluate(MnistReader.TrainImagesPath, MnistReader.TrainLabelsPath);
// evaluate on test data
var testingMetric = neuralnet.evaluate(MnistReader.TestImagesPath, MnistReader.TestLabelsPath);
result.Add(Tuple.Create(epoch, learning_rate, trainingMetric, testingMetric));
string[] row = new string[] { epoch.ToString(), learning_rate.ToString(), trainingMetric.Item1.ToString(), testingMetric.Item1.ToString(),
trainingMetric.Item2.ToString(), testingMetric.Item2.ToString() };
Console.WriteLine(string.Format(format, row));
Utils.writeCSv(result_file, result);
}
}
}
private void btnTrain_Click(object sender, EventArgs e)
{
settings = checkSettings();
System.Threading.Thread t = new System.Threading.Thread(() =>
{
NeuralNetwork n = null;
System.Threading.Thread myThread = new System.Threading.Thread(() =>
{
System.Threading.Thread.CurrentThread.CurrentCulture = new System.Globalization.CultureInfo("en-US");
n = NeuralNetwork.Train(settings);
});
myThread.Start();
myThread.Join();
myNeuralNetwork1 = n;
isTraining = false;
});
if (settings.containsErrors)
{
}
else if (!isTraining)
{
this.progressBar1.Maximum = settings.MaxGens + 1;
this.progressBar1.Visible = true;
t.Name = "NetTrain";
t.Start();
isTraining = true;
btnTrain.Text = "Stop";
}
else if (isTraining)
{
NeuralNetwork.Interrupt();
btnTrain.Text = "Train!";
}
}
