C# Класс Encog.Engine.Opencl.Kernels.KernelNetworkTrain

An OpenCL kernel that is designed to calculate gradients and help train a neural network.
Наследование: EncogKernel
Показать файл Открыть проект Примеры использования класса

Открытые методы

Метод Описание
AssignWorkgroupSizes ( int trainingSize, int requestedGlobalSize ) : void

Assign the workgroup sizes based on the training set size.

Calculate ( int start, int size, bool learn, int iterations ) : void

Calculate one iteration over the specified range.

Compile ( String>.IDictionary options, OpenCLTrainingProfile profile, FlatNetwork network ) : void

Compile the kernel.

Init ( OpenCLTrainingProfile profile ) : void

Setup the kernel.

KernelNetworkTrain ( EncogCLDevice device, FlatNetwork flat, IEngineIndexableSet training, int tempDataSize ) : System

Construct a kernel to train the network.

Release ( ) : void

Release the kernel and all buffers.

Описание методов

AssignWorkgroupSizes() публичный Метод

Assign the workgroup sizes based on the training set size.
public AssignWorkgroupSizes ( int trainingSize, int requestedGlobalSize ) : void
trainingSize int The training set size.
requestedGlobalSize int The requested global size.
Результат void

Calculate() публичный Метод

Calculate one iteration over the specified range.
public Calculate ( int start, int size, bool learn, int iterations ) : void
start int The starting position to calculate for.
size int The ending position to calculate for.
learn bool True, if we should learn.
iterations int The number of iterations to execute.
Результат void

Compile() публичный Метод

Compile the kernel.
public Compile ( String>.IDictionary options, OpenCLTrainingProfile profile, FlatNetwork network ) : void
options String>.IDictionary The options.
profile Encog.Engine.Network.Train.Prop.OpenCLTrainingProfile The OpenCL training profile to use.
network Encog.Engine.Network.Flat.FlatNetwork The network to compile for.
Результат void

Init() публичный Метод

Setup the kernel.
public Init ( OpenCLTrainingProfile profile ) : void
profile Encog.Engine.Network.Train.Prop.OpenCLTrainingProfile
Результат void

KernelNetworkTrain() публичный Метод

Construct a kernel to train the network.
public KernelNetworkTrain ( EncogCLDevice device, FlatNetwork flat, IEngineIndexableSet training, int tempDataSize ) : System
device EncogCLDevice The OpenCL device to use.
flat Encog.Engine.Network.Flat.FlatNetwork The network to train.
training IEngineIndexableSet The training data.
tempDataSize int How much temp data.
Результат System

Release() публичный Метод

Release the kernel and all buffers.
public Release ( ) : void
Результат void