static void testNN()
{
double percentCorrect = 0.0;
int amtLines = 1;
int numCorrect = 0;
StreamReader test = new StreamReader("optdigits_test.txt");
do
{
string testLine = test.ReadLine();
List <int> inputs = testLine.Split(',').Select(x => int.Parse(x)).ToList();
int targetValue = inputs.Last();
NN.forwardPropagate(inputs);
int outputvalue = 0;
int outputLayer_Index = NN.layers.Count - 1;
Node best_Guess_Node = NN.layers[outputLayer_Index].layerNodeArray[0];
Layer outputLayer = NN.layers[outputLayer_Index];
for (int i = 0; i < outputLayer.nodeCount; i++)
{
if (outputLayer.layerNodeArray[i].value > best_Guess_Node.value)
{
best_Guess_Node = outputLayer.layerNodeArray[i];
outputvalue = i;
}
NN.layers[outputLayer_Index].layerNodeArray[i].inputError = (NN.sigmoidDerivate_Func(i - targetValue)) * NN.learningRate;
}
if (outputvalue == targetValue)
{
numCorrect++;
}
amtLines++;
} while (!test.EndOfStream);
percentCorrect = (double)numCorrect / (double)amtLines;
Console.Out.WriteLine(percentCorrect + "");
testResults.Add(percentCorrect + "");
}
static void Main(string[] args)
{
StreamReader reader = new StreamReader("optdigits_train.txt");
NN.generateNN(64);
testResults.Add("% Correct Each 100 Generations");
do
{
for (int runs = 0; runs < 100; runs++)
{
do
{
string line = reader.ReadLine();
List <int> inputs = line.Split(',').Select(x => int.Parse(x)).ToList();
int targetValue = inputs.Last();
NN.forwardPropagate(inputs);
int outputvalue = 0;
int outputLayer_Index = NN.layers.Count - 1;
Node best_Guess_Node = NN.layers[outputLayer_Index].layerNodeArray[0];
Layer outputLayer = NN.layers[outputLayer_Index];
double errorTotal = 0.0;
//Console.WriteLine("____________________________________");
//Console.WriteLine("On line: {0}", lineTag);
//lineTag++;
for (int i = 0; i < outputLayer.nodeCount; i++)
{
if (outputLayer.layerNodeArray[i].value > best_Guess_Node.value)
{
best_Guess_Node = outputLayer.layerNodeArray[i];
outputvalue = i;
}
NN.layers[outputLayer_Index].layerNodeArray[i].inputError = (NN.sigmoidDerivate_Func(i - targetValue)) * NN.learningRate;
//Console.Out.WriteLine(outputLayer.layerNodeArray[i].output);
}
NN.backwardPropagate();
} while (!reader.EndOfStream);
reader.Close();
reader = new StreamReader("optdigits_train.txt");
if (runs % 10 == 0 && runs > 1)
{
testNN();
}
}
} while (Console.ReadLine() != ".");
Console.ReadLine();
StreamWriter writer = new StreamWriter(fileSave);
foreach (string save in testResults)
{
writer.WriteLine(save);
}
writer.Close();
reader.Close();
}