private static void Example()
{
NumberImage[] number = new NumberImage[50];
int countOfCorrectlyRecognized = 0;
LoadAndSaveData.GetInstance().LoadNS(NN);
for (int order = 0; order < 5; order++)
{
for (int num = 0; num < 10; num++)
{
number[order * 10 + num] = LoadAndSaveData.GetInstance().LoadNumberImage(num, order);
int answer = NN.GetAnswer(number[order * 10 + num].GetPixels());
Console.WriteLine(order + "" + num + ".png = " + answer);
if (num == answer)
{
countOfCorrectlyRecognized++;
}
}
Console.WriteLine();
}
Console.WriteLine((double)countOfCorrectlyRecognized / 50 * 100 + "%\n");
}
static void Main(string[] args)
{
LoadAndSaveData dataWork = LoadAndSaveData.GetInstance();
NN = new NeuralNetwork(new int[] { 28 * 28, 500, 150, 10 });
string pixelFile = "train-images.idx3-ubyte";
string labelFile = "train-labels.idx1-ubyte";
string pixelTest = "t10k-images.idx3-ubyte";
string labelTest = "t10k-labels.idx1-ubyte";
input = dataWork.LoadDataMNIST(pixelFile, labelFile, 60000);
test = dataWork.LoadDataMNIST(pixelTest, labelTest, 10000);
correctOutput = new double[60000][];
for (int i = 0; i < 60000; i++)
{
correctOutput[i] = new double[10];
}
for (int i = 0; i < 60000; i++)
{
for (int j = 0; j < 10; j++)
{
correctOutput[i][j] = 0;
correctOutput[i][input[i].GetLabel()] = 1;
}
}
// StartStudy();
Example(); // Example of the work of a neural network on real data (scanned images of numbers)
DigitRecognitionMNIST();
}