org.knime.base.data.neural
Class Perceptron

java.lang.Object
  org.knime.base.data.neural.Perceptron

Direct Known Subclasses:

InputPerceptron, SigmoidPerceptron

public abstract class Perceptron
extends Object

Abstract class defining the behaviour of a perceptron in a neural network.

Author:

Nicolas Cebron, University of Konstanz

Constructor Summary

Perceptron()
Construct an empty perceptron (cannot be used properly unless the weights are set).

Perceptron(double[] weights)
Construct a perceptron with given weights.

Perceptron(int nrInputs)
Construct a perceptron with given number of weights.

Method Summary

abstract double	activationFunction(double in) Activation function.
double	evaluate(double[] in) Evaluates a given input for the perceptron.
String	getClassValue()
double	getThreshold() Returns the threshold.
double	getWeight(int i) Get weight at given position.
double[]	getWeights()
abstract double	output() Used to get the output from all input neurons.
static double	random() Returns a randomly chosen double between -1 and 1.
static double	scalarProduct(double[] x, double[] y) Computes the scalar Product of two vectors, represented as double arrays.
void	setClassValue(String classval) Allows to set a class value for a perceptron (optional).
void	setThreshold(double threshold) Sets the threshold.
void	setWeight(int i, double weight) Sets weight at given position.
void	setWeights(double[] weights) Set all new weights.

Methods inherited from class java.lang.Object

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

Constructor Detail

Perceptron

public Perceptron(int nrInputs)

Construct a perceptron with given number of weights. Weights and threshold are randomly initialized.

Parameters:

nrInputs - number of inputs for the new perceptron

Perceptron

public Perceptron()

Construct an empty perceptron (cannot be used properly unless the weights are set).

Perceptron

public Perceptron(double[] weights)

Construct a perceptron with given weights.

Parameters:

weights - weights for the new perceptron

Method Detail

getWeights

public double[] getWeights()

Returns:

weights of the perceptron

getWeight

public double getWeight(int i)

Get weight at given position.

Parameters:

i - position

Returns:

weight at position i

setWeights

public void setWeights(double[] weights)

Set all new weights.

Parameters:

weights - new weights to set

setWeight

public void setWeight(int i,
                      double weight)

Sets weight at given position.

Parameters:

i - position

weight - new value for weight at position i

activationFunction

public abstract double activationFunction(double in)

Activation function. Implement it to create the desired behaviour for the perceptron.

Parameters:

in - input value

Returns:

output value for activation

evaluate

public double evaluate(double[] in)

Evaluates a given input for the perceptron.

Parameters:

in - input to evaluate

Returns:

activation function applied to the weighted sum of the inputs

output

public abstract double output()

Used to get the output from all input neurons.

Returns:

activation function applied to the weighted sum of all inputs, which are taken from the input neurons

getThreshold

public double getThreshold()

Returns the threshold.

Returns:

threshold of the neutron

setThreshold

public void setThreshold(double threshold)

Sets the threshold.

Parameters:

threshold - The threshold to set

setClassValue

public void setClassValue(String classval)

Allows to set a class value for a perceptron (optional).

Parameters:

classval - the class value to set.

getClassValue

public String getClassValue()

Returns:

class value of the perceptron if it exists.

random

public static double random()

Returns a randomly chosen double between -1 and 1.

Returns:

random double number between -1 and 1

scalarProduct

public static double scalarProduct(double[] x,
                                   double[] y)

Computes the scalar Product of two vectors, represented as double arrays.

Parameters:

x - first array

y - second array

Returns:

scalar product of x and y