C# Класс NNX.Core.Training.UntilDoneGradientTrainer

An on-line gradient descent trainer that radnomly splits the input trainingSet into a training and validation subsets. The trainier will keep selecting a random input/target pair from the training set and applying backpropagation until cross-entropy error on the validation set does not improve after MaxEpochsWithoutImprovement. The trainer will terminate after NumEpochs even if MaxEpochsWithoutImprovement condition has not been reached.
Наследование: SimpleGradientTrainer
Показать файл Открыть проект Примеры использования класса

Открытые методы

Метод Описание
GetValidationSetFraction ( ) : double
Train ( IList trainingSet, IList validationSet, IRandomGenerator rand, INeuralNetwork nn ) : void
Validate ( ) : void

Описание методов

GetValidationSetFraction() публичный Метод

public GetValidationSetFraction ( ) : double
Результат double

Train() публичный Метод

public Train ( IList trainingSet, IList validationSet, IRandomGenerator rand, INeuralNetwork nn ) : void
trainingSet IList
validationSet IList
rand IRandomGenerator
nn INeuralNetwork
Результат void

Validate() публичный Метод

public Validate ( ) : void
Результат void