private static void CreateNetAndTreainer(ReadoutLayerSettings.ReadoutUnitSettings settings,
List <double[]> trainingPredictorsCollection,
List <double[]> trainingIdealOutputsCollection,
System.Random rand,
out INonRecurrentNetwork net,
out INonRecurrentNetworkTrainer trainer
)
{
if (settings.NetType == ReadoutLayerSettings.ReadoutUnitSettings.ReadoutUnitNetworkType.FF)
{
FeedForwardNetworkSettings netCfg = (FeedForwardNetworkSettings)settings.NetSettings;
FeedForwardNetwork ffn = new FeedForwardNetwork(trainingPredictorsCollection[0].Length, 1, netCfg);
net = ffn;
switch (netCfg.RegressionMethod)
{
case FeedForwardNetworkSettings.TrainingMethodType.Linear:
trainer = new LinRegrTrainer(ffn, trainingPredictorsCollection, trainingIdealOutputsCollection, settings.RegressionAttemptEpochs, rand, netCfg.LinRegrTrainerCfg);
break;
case FeedForwardNetworkSettings.TrainingMethodType.Resilient:
trainer = new RPropTrainer(ffn, trainingPredictorsCollection, trainingIdealOutputsCollection, netCfg.RPropTrainerCfg);
break;
default:
throw new ArgumentException($"Not supported regression method {netCfg.RegressionMethod}");
}
}
else
{
ParallelPerceptronSettings netCfg = (ParallelPerceptronSettings)settings.NetSettings;
ParallelPerceptron ppn = new ParallelPerceptron(trainingPredictorsCollection[0].Length, netCfg);
net = ppn;
trainer = new PDeltaRuleTrainer(ppn, trainingPredictorsCollection, trainingIdealOutputsCollection, netCfg.PDeltaRuleTrainerCfg);
}
net.RandomizeWeights(rand);
return;
}
/// <summary>
/// Prepares trained readout unit for specified output field and task.
/// </summary>
/// <param name="taskType">Type of the task</param>
/// <param name="readoutUnitIdx">Index of the readout unit (informative only)</param>
/// <param name="outputFieldName">Name of the corresponding output field (informative only)</param>
/// <param name="foldNum">Current fold number</param>
/// <param name="numOfFolds">Total number of the folds</param>
/// <param name="refBinDistr">Reference bin distribution (if task type is Classification)</param>
/// <param name="trainingPredictorsCollection">Collection of the predictors for training</param>
/// <param name="trainingIdealOutputsCollection">Collection of ideal outputs for training. Note that the double array always has only one member.</param>
/// <param name="testingPredictorsCollection">Collection of the predictors for testing</param>
/// <param name="testingIdealOutputsCollection">Collection of ideal outputs for testing. Note that the double array always has only one member.</param>
/// <param name="rand">Random object to be used</param>
/// <param name="readoutUnitSettings">Readout unit configuration parameters</param>
/// <param name="controller">Regression controller</param>
/// <param name="controllerUserObject">An user object to be passed to controller</param>
/// <returns>Prepared readout unit</returns>
public static ReadoutUnit CreateTrained(CommonEnums.TaskType taskType,
int readoutUnitIdx,
string outputFieldName,
int foldNum,
int numOfFolds,
BinDistribution refBinDistr,
List <double[]> trainingPredictorsCollection,
List <double[]> trainingIdealOutputsCollection,
List <double[]> testingPredictorsCollection,
List <double[]> testingIdealOutputsCollection,
System.Random rand,
ReadoutLayerSettings.ReadoutUnitSettings readoutUnitSettings,
RegressionCallbackDelegate controller = null,
Object controllerUserObject = null
)
{
ReadoutUnit bestReadoutUnit = new ReadoutUnit();
//Regression attempts
bool stopRegression = false;
for (int regrAttemptNumber = 1; regrAttemptNumber <= readoutUnitSettings.RegressionAttempts; regrAttemptNumber++)
{
//Create network and trainer
INonRecurrentNetwork net;
INonRecurrentNetworkTrainer trainer;
CreateNetAndTreainer(readoutUnitSettings,
trainingPredictorsCollection,
trainingIdealOutputsCollection,
rand,
out net,
out trainer
);
//Reference binary distribution
//Iterate training cycles
for (int epoch = 1; epoch <= readoutUnitSettings.RegressionAttemptEpochs; epoch++)
{
trainer.Iteration();
List <double[]> trainingComputedOutputsCollection = null;
List <double[]> testingComputedOutputsCollection = null;
//Compute current error statistics after training iteration
ReadoutUnit currReadoutUnit = new ReadoutUnit();
currReadoutUnit.Network = net;
currReadoutUnit.TrainingErrorStat = net.ComputeBatchErrorStat(trainingPredictorsCollection, trainingIdealOutputsCollection, out trainingComputedOutputsCollection);
if (taskType == CommonEnums.TaskType.Classification)
{
currReadoutUnit.TrainingBinErrorStat = new BinErrStat(refBinDistr, trainingComputedOutputsCollection, trainingIdealOutputsCollection);
currReadoutUnit.CombinedBinaryError = currReadoutUnit.TrainingBinErrorStat.TotalErrStat.Sum;
//currReadoutUnit.CombinedBinaryError = currReadoutUnit.TrainingBinErrorStat.ProportionalErr;
}
currReadoutUnit.CombinedPrecisionError = currReadoutUnit.TrainingErrorStat.ArithAvg;
if (testingPredictorsCollection != null && testingPredictorsCollection.Count > 0)
{
currReadoutUnit.TestingErrorStat = net.ComputeBatchErrorStat(testingPredictorsCollection, testingIdealOutputsCollection, out testingComputedOutputsCollection);
currReadoutUnit.CombinedPrecisionError = Math.Max(currReadoutUnit.CombinedPrecisionError, currReadoutUnit.TestingErrorStat.ArithAvg);
if (taskType == CommonEnums.TaskType.Classification)
{
currReadoutUnit.TestingBinErrorStat = new BinErrStat(refBinDistr, testingComputedOutputsCollection, testingIdealOutputsCollection);
currReadoutUnit.CombinedBinaryError = Math.Max(currReadoutUnit.CombinedBinaryError, currReadoutUnit.TestingBinErrorStat.TotalErrStat.Sum);
//currReadoutUnit.CombinedBinaryError = Math.Max(currReadoutUnit.CombinedBinaryError, currReadoutUnit.TestingBinErrorStat.ProportionalErr);
}
}
//Current results processing
bool better = false, stopTrainingCycle = false;
//Result first initialization
if (bestReadoutUnit.CombinedPrecisionError == -1)
{
//Adopt current regression results
bestReadoutUnit = currReadoutUnit.DeepClone();
}
//Perform call back if it is defined
RegressionControlOutArgs cbOut = null;
if (controller != null)
{
//Evaluation of the improvement is driven externaly
RegressionControlInArgs cbIn = new RegressionControlInArgs();
cbIn.TaskType = taskType;
cbIn.ReadoutUnitIdx = readoutUnitIdx;
cbIn.OutputFieldName = outputFieldName;
cbIn.FoldNum = foldNum;
cbIn.NumOfFolds = numOfFolds;
cbIn.RegrAttemptNumber = regrAttemptNumber;
cbIn.RegrMaxAttempts = readoutUnitSettings.RegressionAttempts;
cbIn.Epoch = epoch;
cbIn.MaxEpochs = readoutUnitSettings.RegressionAttemptEpochs;
cbIn.TrainingPredictorsCollection = trainingPredictorsCollection;
cbIn.TrainingIdealOutputsCollection = trainingIdealOutputsCollection;
cbIn.TrainingComputedOutputsCollection = trainingComputedOutputsCollection;
cbIn.TestingPredictorsCollection = testingPredictorsCollection;
cbIn.TestingIdealOutputsCollection = testingIdealOutputsCollection;
cbIn.TestingComputedOutputsCollection = testingComputedOutputsCollection;
cbIn.CurrReadoutUnit = currReadoutUnit;
cbIn.BestReadoutUnit = bestReadoutUnit;
cbIn.UserObject = controllerUserObject;
cbOut = controller(cbIn);
better = cbOut.CurrentIsBetter;
stopTrainingCycle = cbOut.StopCurrentAttempt;
stopRegression = cbOut.StopRegression;
}
else
{
//Default implementation
better = IsBetter(taskType, currReadoutUnit, bestReadoutUnit);
}
//Best?
if (better)
{
//Adopt current regression results
bestReadoutUnit = currReadoutUnit.DeepClone();
}
//Training stop conditions
if (stopTrainingCycle || stopRegression)
{
break;
}
}
//Regression stop conditions
if (stopRegression)
{
break;
}
}
//Create statistics of the best network weights
bestReadoutUnit.OutputWeightsStat = bestReadoutUnit.Network.ComputeWeightsStat();
return(bestReadoutUnit);
}
