private void Load(string filePath)
{
FileInfo fileInfo = new FileInfo(filePath);
if (fileInfo.Exists)
{
Stream stream = File.Open(filePath, FileMode.Open, FileAccess.Read);
BinaryFormatter formatter = new BinaryFormatter();
network = (CNNetwork)formatter.Deserialize(stream);
stream.Close();
stream = null;
formatter = null;
}
}
private void Train()
{
int outputNodes = 10;
double learningRate = 0.1;
int bias = 1;
CNNetwork.ECNNLayer[] layers = new CNNetwork.ECNNLayer[]
{
CNNetwork.ECNNLayer.ConvLayer,
CNNetwork.ECNNLayer.PoolingLayer,
CNNetwork.ECNNLayer.ConvLayer,
CNNetwork.ECNNLayer.PoolingLayer,
//CNNetwork.ECNNLayer.ConvLayer
};
int[] strides = new int[]
{
1, 1, 1
};
Size[] sizes = new Size[]
{
new Size(3, 3),
new Size(3, 3),
new Size(3, 3)
};
int[] kernelsCount = new int[]
{
4, 4
};
int[] layersNode = new int[]
{
100,
10
};
if (network == null)
{
network = new CNNetwork(new Size(30, 30), layers.Length,
layers, strides, sizes, kernelsCount, layersNode, learningRate, bias);
Console.WriteLine("Not exist");
Console.ReadKey();
}
else
{
Console.WriteLine("Exist");
Console.ReadKey();
}
int count = 0;
for (int appch = 0; appch < 1; appch++)
{
using (FileStream file = new FileStream("mnist_train.csv", FileMode.Open, FileAccess.Read))
using (StreamReader reader = new StreamReader(file))
{
string record = null;
while ((record = reader.ReadLine()) != null)
{
string[] all_values = record.Split(',');
double[] inputs = new double[all_values.Length - 1];
for (int i = 1; i < all_values.Length; i++)
{
inputs[i - 1] = (double.Parse(all_values[i]) / 255 * 0.99) + 0.01;
}
double[] targets = new double[outputNodes];
for (int i = 0; i < outputNodes; i++)
{
targets[i] = 0.01;
}
targets[int.Parse(all_values[0])] = 0.99;
Mat <double> image = new Mat <double>(28, 28, inputs);
image = image.Padding(1);
double error = network.Train(image, targets);
Console.WriteLine("[{0}] Sum of Error : {1}", ++count, error);
if (count % 10000 == 0)
{
Save("CNN_" + count.ToString() + ".info");
}
}
}
}
}
