public static void Train()
{
var holder = new CsvHolder();
holder.LoadTrainingSet();
var nn = new NeuralNet();
var batches = 1875;
var batchesSize = 32;
var epochs = 100;
for (int i = 0; i < epochs; i++)
{
var loss = 0.0;
Console.WriteLine("EPOCH {0} OF {1}", i + 1, epochs);
for (int j = 0; j < batches; j++)
{
string xFilter = j * batchesSize + ":" + (j + 1) * batchesSize;
string filter = xFilter + ",1:";
string filter2 = xFilter + ",0";
if (j % 75 == 0)
{
Console.WriteLine("[EPOCH {0} - BATCH {1} OF {2}] :: Using filters \"{3}\" and \"{4}\"", i + 1, j, batches, filter, filter2);
}
var dataX = (holder.TrainingSet[filter].T) / 255;
var dataY = holder.TrainingSet[filter2].T;
loss += nn.Train(dataX, dataY);
}
Console.WriteLine("EPOCH {0} DONE. AVERAGE LOSS = {1}", i, loss / batches);
Shuffle(holder.TrainingSet);
nn.PrintWB();
}
}
public Controller()
{
_nn = new NeuralNet.NeuralNet();
_holder = new CsvHolder();
_holder.LoadTestSet();
}
