Package libai.nn.supervised

Class RBF

java.lang.Object

libai.nn.NeuralNetwork

libai.nn.supervised.SupervisedLearning

libai.nn.supervised.Perceptron

libai.nn.supervised.Adaline

libai.nn.supervised.RBF

All Implemented Interfaces:

java.io.Serializable

public class RBF
extends Adaline

Radial Basis Function or RBF. Is an hybrid neural network with 3 layers (1-input, 1-hidden, 1-output). The hidden layers is trained using a stochastic clustering algorithm: k-means. The final layer is trained using the Adaline rule. The k-means algorithm is used to set up the position of the "centers" of the radial basis functions, as this process is regardless of the output and invariant over the input, could be used a highly efficient algorithm. This implementation uses only Gaussian functions as radial basis functions.

See Also:

Serialized Form

Field Summary

Fields

Modifier and Type	Field	Description
protected int[]	nperlayer
protected double[]	sigma

Fields inherited from class libai.nn.NeuralNetwork

plotter, progress, random

Fields inherited from class libai.nn.supervised.Perceptron

b, ins, outs, signum, W

Constructor Summary

Constructors

Constructor	Description
RBF(int[] nperlayer)	Constructor.
RBF(int[] nperlayer, java.util.Random rand)	Constructor.

Method Summary

Modifier and Type	Method	Description
Column	simulate(Column pattern)	Calculates the output for the pattern.
void	simulate(Column pattern, Column result)	Calculate the output for the pattern and left the result on result.
void	train(Column[] patterns, Column[] answers, double alpha, int epochs, int offset, int length, double minerror)	Train the network using a hybrid scheme.

Methods inherited from class libai.nn.NeuralNetwork

error, error, euclideanDistance2, euclideanDistance2, gaussian, getDefaultRandomGenerator, getPlotter, getProgressBar, initializeProgressBar, open, open, open, save, setPlotter, setProgressBar, train, train

Methods inherited from class java.lang.Object

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

Methods inherited from class libai.nn.supervised.Perceptron

getWeights

Methods inherited from class libai.nn.supervised.SupervisedLearning

validatePreconditions

Field Detail

nperlayer

protected final int[] nperlayer

sigma

protected final double[] sigma

Constructor Detail

RBF

public RBF(int[] nperlayer)

Constructor. Receives an array with the information of the number of neurons per layer. Layer[0] is the input layer. Layer[1] is the hidden layer and represents the number radial functions to use. layer[2] is the output layer.

Parameters:

nperlayer - Neurons Per Layer.

RBF

public RBF(int[] nperlayer,
           java.util.Random rand)

Constructor. Receives an array with the information of the number of neurons per layer. Layer[0] is the input layer. Layer[1] is the hidden layer and represents the number radial functions to use. layer[2] is the output layer.

Parameters:

nperlayer - Neurons Per Layer.

rand - Random generator used for creating matrices

Method Detail

train

public void train(Column[] patterns,
                  Column[] answers,
                  double alpha,
                  int epochs,
                  int offset,
                  int length,
                  double minerror)

Train the network using a hybrid scheme. First set the centers of the radial basis functions using k-means algorithm. After that the radius of the function is calculated using n-nearest neighbors. When n = the number of inputs. Then the output for the hidden layer are precalculated and used as input for the Adaline training.

Overrides:

train in class Adaline

Parameters:

patterns - The patterns to be learned.

answers - The expected answers.

alpha - The learning rate.

epochs - The maximum number of iterations

offset - The first pattern position

length - How many patterns will be used.

minerror - The minimal error expected.

simulate

public Column simulate(Column pattern)

Description copied from class: NeuralNetwork

Calculates the output for the pattern.

Overrides:

simulate in class Perceptron

Parameters:

pattern - Pattern to use as input.

Returns:

The output for the neural network.

simulate

public void simulate(Column pattern,
                     Column result)

Description copied from class: Adaline

Calculate the output for the pattern and left the result on result. result = W * pattern + b

Overrides:

simulate in class Adaline

Parameters:

pattern - The input pattern

result - The output result.