C# Класс Encog.Engine.Network.Flat.FlatNetwork

Implements a flat (vector based) neural network in the Encog Engine. This is meant to be a very highly efficient feedforward, or simple recurrent, neural network. It uses a minimum of objects and is designed with one principal in mind-- SPEED. Readability, c reuse, object oriented programming are all secondary in consideration. Vector based neural networks are also very good for GPU processing. The flat network classes will make use of the GPU if you have enabled GPU processing. See the Encog class for more info.

Наследование: IEngineNeuralNetwork

Показать файл Открыть проект Примеры использования класса

Открытые методы

Метод	Описание
CalculateError ( IEngineIndexableSet data ) : double	Calculate the error for this neural network. The error is calculated using root-mean-square(RMS).
ClearConnectionLimit ( ) : void	Clear the connection limit.
ClearContext ( ) : void	Clear any context neurons.
Clone ( ) : Object	Clone the network.
CloneFlatNetwork ( FlatNetwork result ) : void	Clone a flat network.
Compute ( double input, double output ) : void	Calculate the output for the given input.
DecodeNetwork ( double data ) : void	Dec the specified data into the weights of the neural network. This method performs the opposite of encNetwork.
EncodeNetwork ( ) : double[]	Enc the neural network to an array of doubles. This includes the network weights. To read this into a neural network, use the decNetwork method.
FlatNetwork ( ) : Encog.Engine	Default constructor.
FlatNetwork ( FlatLayer layers ) : Encog.Engine	Create a flat network from an array of layers.
FlatNetwork ( int input, int hidden1, int hidden2, int output, bool tanh ) : Encog.Engine	Construct a flat neural network.
HasSameActivationFunction ( ) : Type	Neural networks with only one type of activation function offer certain optimization options. This method determines if only a single activation function is used.
Init ( FlatLayer layers ) : void	Construct a flat network.
Randomize ( ) : void	Perform a simple randomization of the weights of the neural network between -1 and 1.
Randomize ( double hi, double lo ) : void	Perform a simple randomization of the weights of the neural network between the specified hi and lo.

Защищенные методы

Метод	Описание
ComputeLayer ( int currentLayer ) : void	Calculate a layer.

Описание методов

CalculateError() публичный Метод

Calculate the error for this neural network. The error is calculated using root-mean-square(RMS).

public CalculateError ( IEngineIndexableSet data ) : double
data	IEngineIndexableSet	The training set.
Результат	double

ClearConnectionLimit() публичный Метод

Clear the connection limit.

public ClearConnectionLimit ( ) : void
Результат	void

ClearContext() публичный Метод

Clear any context neurons.

public ClearContext ( ) : void
Результат	void

Clone() публичный Метод

Clone the network.

public Clone ( ) : Object
Результат	Object

CloneFlatNetwork() публичный Метод

Clone a flat network.

public CloneFlatNetwork ( FlatNetwork result ) : void
result	FlatNetwork	The cloned flat network.
Результат	void

Compute() публичный Метод

Calculate the output for the given input.

public Compute ( double input, double output ) : void
input	double	The input.
output	double	Output will be placed here.
Результат	void

ComputeLayer() защищенный Метод

Calculate a layer.

protected ComputeLayer ( int currentLayer ) : void
currentLayer	int	The layer to calculate.
Результат	void

DecodeNetwork() публичный Метод

Dec the specified data into the weights of the neural network. This method performs the opposite of encNetwork.

public DecodeNetwork ( double data ) : void
data	double	The data to be decd.
Результат	void

EncodeNetwork() публичный Метод

Enc the neural network to an array of doubles. This includes the network weights. To read this into a neural network, use the decNetwork method.

public EncodeNetwork ( ) : double[]
Результат	double[]

FlatNetwork() публичный Метод

Default constructor.

public FlatNetwork ( ) : Encog.Engine
Результат	Encog.Engine

FlatNetwork() публичный Метод

Create a flat network from an array of layers.

public FlatNetwork ( FlatLayer layers ) : Encog.Engine
layers	FlatLayer	The layers.
Результат	Encog.Engine

FlatNetwork() публичный Метод

Construct a flat neural network.

public FlatNetwork ( int input, int hidden1, int hidden2, int output, bool tanh ) : Encog.Engine
input	int	Neurons in the input layer.
hidden1	int	Neurons in the first hidden layer. Zero for no first hiddenlayer.
hidden2	int	Neurons in the second hidden layer. Zero for no second hiddenlayer.
output	int	Neurons in the output layer.
tanh	bool	True if this is a tanh activation, false for sigmoid.
Результат	Encog.Engine

HasSameActivationFunction() публичный Метод

Neural networks with only one type of activation function offer certain optimization options. This method determines if only a single activation function is used.

public HasSameActivationFunction ( ) : Type
Результат	System.Type

Init() публичный Метод

Construct a flat network.

public Init ( FlatLayer layers ) : void
layers	FlatLayer	The layers of the network to create.
Результат	void

Randomize() публичный Метод

Perform a simple randomization of the weights of the neural network between -1 and 1.

public Randomize ( ) : void
Результат	void

Randomize() публичный Метод

Perform a simple randomization of the weights of the neural network between the specified hi and lo.

public Randomize ( double hi, double lo ) : void
hi	double	The network high.
lo	double	The network low.
Результат	void