private void Button_TestAll_Click(object sender, EventArgs e)
{
double learningRate;
int hiddenlayer;
int count = 0;
int countHit = 0;
int countMiss = 0;
if (double.TryParse(textBox_learningRate.Text, out learningRate) &&
int.TryParse(textBox_hiddenLayer.Text, out hiddenlayer))
{
if (cnn == null)
{
cnn = new MNIST.CNN.CNN(hiddenlayer, allLabel, lengthImage, learningRate);
}
if (mnist == null)
{
mnist = new Reader.MnistReader(textBox_pathTrainFile.Text
, textBox_pathLabelFile.Text
, textBox_pathTestFile.Text
, textBox_pathTestLabelFile.Text);
}
foreach (var image in mnist.ReadTestData())
{
int labelDataSet = image.Label;
double[] oneHotEncoding = convertLabelToOneHotEncoding(labelDataSet, allLabel.Length);
byte[,] images = image.Data;
int[] flat = Convert2ArrayTo1Array(images);
double[] flatDouble = flat.Select(Convert.ToDouble).ToArray();
Matrix <double> flatMatrix = cnn.convertArrayToMatrix(flatDouble);
Matrix <double> pOutputMatrix = cnn.predict(flatMatrix);
string answer = cnn.answer(pOutputMatrix);
if (labelDataSet.ToString().Trim() == answer.Trim())
{
countHit++;
}
else
{
countMiss++;
}
count++;
}
mnist.closeRead();
MessageBox.Show(string.Format("Total: {0} Hit: {1} Miss: {2}", count, countHit, countMiss));
}
}
private void Button_testOneImage_Click(object sender, EventArgs e)
{
double learningRate;
int hiddenlayer;
int count = 0;
if (double.TryParse(textBox_learningRate.Text, out learningRate) &&
int.TryParse(textBox_hiddenLayer.Text, out hiddenlayer))
{
if (cnn == null)
{
cnn = new MNIST.CNN.CNN(hiddenlayer, allLabel, lengthImage, learningRate);
}
if (mnist == null)
{
mnist = new Reader.MnistReader(textBox_pathTrainFile.Text
, textBox_pathLabelFile.Text
, textBox_pathTestFile.Text
, textBox_pathTestLabelFile.Text);
}
foreach (var image in mnist.ReadTestData())
{
if (count == countStepTest)
{
int labelDataSet = image.Label;
double[] oneHotEncoding = convertLabelToOneHotEncoding(labelDataSet, allLabel.Length);
byte[,] images = image.Data;
int[] flat = Convert2ArrayTo1Array(images);
double[] flatDouble = nomalizeData(flat);
//double[] flatDouble = flat.Select(Convert.ToDouble).ToArray();
Bitmap bmp;
if (convertFlatToImage(flat, 28, 28, out bmp))
{
pictureBox_draw.Image = bmp;
}
Matrix <double> flatMatrix = cnn.convertArrayToMatrix(flatDouble);
Matrix <double> pOutputMatrix = cnn.predict(flatMatrix);
string answer = cnn.answer(pOutputMatrix);
label_target.Text = answer;
countStepTest = (countStepTest + 1) % 1000;
break;
}
count++;
}
mnist.closeRead();
}
}
private void Button_train_Click(object sender, EventArgs e)
{
double learningRate;
int hiddenlayer;
if (double.TryParse(textBox_learningRate.Text, out learningRate) &&
int.TryParse(textBox_hiddenLayer.Text, out hiddenlayer))
{
if (cnn == null)
{
cnn = new MNIST.CNN.CNN(hiddenlayer, allLabel, lengthImage, learningRate);
}
if (mnist == null)
{
mnist = new Reader.MnistReader(textBox_pathTrainFile.Text
, textBox_pathLabelFile.Text
, textBox_pathTestFile.Text
, textBox_pathTestLabelFile.Text);
}
for (int numTrain = 0; numTrain <= 100; numTrain++)
{
foreach (var image in mnist.ReadTrainingData())
{
int labelDataSet = image.Label;
double[] oneHotEncoding = convertLabelToOneHotEncoding(labelDataSet, allLabel.Length);
byte[,] images = image.Data;
int[] flat = Convert2ArrayTo1Array(images);
double[] flatDouble = nomalizeData(flat);
//double[] flatDouble = flat.Select(Convert.ToDouble).ToArray();
cnn.train(flatDouble, oneHotEncoding);
trainCount++;
//MessageBox.Show(answer);
#region test draw digit
////8 digit
//int[] flat = new int[] {0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,43,47,47,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,9,108,249,253,253,208,207,207,207,149,65,13,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,9,184,254,253,253,253,254,253,253,253,254,253,213,25,0,0,0,0,0,0,0,0,0,0,0,0,0,55,203,254,254,199,127,127,60,93,84,68,151,222,254,161,0,0,0,0,0,0,0,0,0,0,0,0,0,138,253,253,199,19,0,0,0,0,0,0,0,155,253,211,0,0,0,0,0,0,0,0,0,0,0,0,0,138,253,253,17,0,0,0,0,0,0,0,74,241,253,211,0,0,0,0,0,0,0,0,0,0,0,0,0,105,253,253,102,0,0,0,0,0,0,34,229,253,253,160,0,0,0,0,0,0,0,0,0,0,0,0,0,0,149,254,229,40,0,0,0,38,153,254,254,254,180,25,0,0,0,0,0,0,0,0,0,0,0,0,0,0,19,198,254,207,9,34,72,235,253,253,224,139,13,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,17,211,253,215,240,254,253,234,128,17,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,102,229,253,253,253,228,77,13,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,70,170,254,254,254,254,254,254,119,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,26,130,230,254,253,253,185,115,64,211,253,248,21,0,0,0,0,0,0,0,0,0,0,0,0,0,0,166,232,253,253,247,162,46,13,7,91,245,253,254,56,0,0,0,0,0,0,0,0,0,0,0,0,0,0,128,253,253,253,210,93,127,159,204,253,253,253,228,15,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,134,241,254,255,254,254,254,254,254,254,228,34,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,27,115,140,206,206,206,207,206,123,15,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0
//};
//int[] flat = convertImageToFlatGrayScale(bitmap28_28);
//Bitmap img;
//if (convertFlatToImage(flat, 28, 28, out img))
//{
// pictureBox_draw.Image = new Bitmap(img, new Size(pictureBox_draw.Width, pictureBox_draw.Height));
//}
#endregion
//break;
}
}
MessageBox.Show("Train finish " + trainCount);
}
}