com.kylecorry.lann

Class NeuralNetwork

java.lang.Object

com.kylecorry.lann.NeuralNetwork

public class NeuralNetwork
extends java.lang.Object

Nested Class Summary

Nested Classes

Modifier and Type	Class and Description
static class	NeuralNetwork.Builder Builder for creating neural network instances.

Method Summary

Modifier and Type	Method and Description
static int	argMax(Matrix output) Get the position of the most probable in an output array.
double	crossEntropyError(Matrix x, Matrix y) Calculate the cross entropy error of the neural network.
Matrix	getWeights(int layer) Get the weight matrix of a layer of the neural network.
void	load(java.io.File file) Loads a neural network from a file.
void	load(java.io.InputStream is) Loads a neural network from an InputStream.
void	load(java.lang.String filename) Loads a neural network from a file.
Matrix	predict(Matrix input) Give a prediction based on some input.
void	save(java.io.File file) Saves the neural network to a file (CSV).
void	save(java.lang.String filename) Saves the neural network to a file (CSV).
void	setWeights(int layer, Matrix weights) Set the weight matrix of a layer of the neural network.
int	size() Get the size of the neural network.
double	squaredError(Matrix x, Matrix y) Calculate the squared error of the neural network.
double	train(Matrix input, Matrix output, double learningRate) Train the neural network to predict an output given some input.

Methods inherited from class java.lang.Object

equals, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Method Detail

crossEntropyError

public double crossEntropyError(Matrix x,
                                Matrix y)

Calculate the cross entropy error of the neural network.

Parameters:

x - The input to the neural network.

y - The expected output.

Returns:

The cross entropy error.

squaredError

public double squaredError(Matrix x,
                           Matrix y)

Calculate the squared error of the neural network.

Parameters:

x - The input to the neural network.

y - The expected output.

Returns:

The squared error.

predict

public Matrix predict(Matrix input)

Give a prediction based on some input.

Parameters:

input - The input to the neural network which is equal in size to the number of input neurons.

Returns:

The output of the neural network.

argMax

public static int argMax(Matrix output)

Get the position of the most probable in an output array.

Parameters:

output - The output of the neural network (using Softmax)

Returns:

The position of the most probable class.

getWeights

public Matrix getWeights(int layer)

Get the weight matrix of a layer of the neural network.

Parameters:

layer - The layer number of the neural network.

Returns:

The weights of the corresponding layer.

setWeights

public void setWeights(int layer,
                       Matrix weights)

Set the weight matrix of a layer of the neural network.

Parameters:

layer - The layer number of the neural network.

weights - The new weight matrix for the layer.

size

public int size()

Get the size of the neural network.

Returns:

The number of layers in the neural network.

train

public double train(Matrix input,
                    Matrix output,
                    double learningRate)

Train the neural network to predict an output given some input.

Parameters:

input - The input to the neural network.

output - The target output for the given input.

learningRate - The rate at which the neural network learns. This is normally 0.01.

Returns:

The error of the network as an mean cross entropy.

save

public void save(java.io.File file)

Saves the neural network to a file (CSV).

Parameters:

file - The file in which to save the weights to.

save

public void save(java.lang.String filename)

Saves the neural network to a file (CSV).

Parameters:

filename - The filename in which to save the weights to.

load

public void load(java.io.InputStream is)

Loads a neural network from an InputStream.

Parameters:

is - The InputStream to retrieve the weights from.

load

public void load(java.io.File file)

Loads a neural network from a file.

Parameters:

file - The file to retrieve the weights from.

load

public void load(java.lang.String filename)

Loads a neural network from a file.

Parameters:

filename - The name of the file to retrieve the weights from.