public int StockTest(ref StockDataSet data)
{
data.Reset();
int correct = 0;
int total = data.getSize();
for (int i = 0; i < total; i++)
{
StockDataForNetwork networkData = data.GetNextNetworkData();
SetInputLayer(networkData.InputLayer);
FeedForward();
float predictedValue = mNeurons[lastLayerIndex][0];
if (predictedValue >= 0.5f && networkData.OutputLayer[0] >= 0.5f)
{
correct++;
continue;
}
if (predictedValue < 0.5f && networkData.OutputLayer[0] < 0.5f)
{
correct++;
continue;
}
}
data.Reset();
return(correct);
}
/// <summary>
/// Stochastic gradient descent learning algorithm.
/// </summary>
public void SGD(ref StockDataSet trainingData, ref StockDataSet testingData, int epochs, int batchSize, float learningRate, bool output)
{
MathNet.Numerics.Control.UseSingleThread(); // Single thread is optimal over multithreading.
Console.WriteLine("Training...");
for (int l = 0; l < vSizes.Count; l++)
{
mSumW[l].Clear();
mSumB[l].Clear();
}
RandomizeParameters();
for (int i = 0; i < epochs; i++)
{
StockDataForNetwork networkData = trainingData.GetNextNetworkData();
int j = 0;
while (networkData != null)
{
SetInputLayer(networkData.InputLayer);
FeedForward();
SetOutputLayer(networkData.OutputLayer);
OutputError();
Backprop();
// Perform summation calculations over the training images.
for (int l = lastLayerIndex; l > 0; l--)
{
mSumW[l] += mError[l].ToColumnMatrix() * mNeurons[l - 1].ToRowMatrix();
mSumB[l] += mError[l];
}
// Update the weights and biases after the mini-batch has been processed.
if (j % batchSize == 0 && j > 0)
{
for (int l = 0; l < vSizes.Count; l++)
{
mWeights[l] += mSumW[l].Multiply(-learningRate / batchSize);
mBiases[l] += mSumB[l].Multiply(-learningRate / batchSize);
mSumW[l].Clear();
mSumB[l].Clear();
}
}
networkData = trainingData.GetNextNetworkData();
j++;
}
if (output == true)
{
Console.WriteLine("Epoch {0}: {1} / {2}", i, StockTest(ref testingData), testingData.getSize());
}
trainingData.Reset();
}
Console.WriteLine("Training Complete.");
}