JMSLTM Numerical Library 6.1

com.imsl.datamining.neural
Class Perceptron

java.lang.Object
  extended by com.imsl.datamining.neural.Node
      extended by com.imsl.datamining.neural.Perceptron
All Implemented Interfaces:
Serializable
Direct Known Subclasses:
OutputPerceptron

public class Perceptron
extends Node

A Perceptron node in a neural network. Perceptrons are created by factory methods in a Network Layer.

Each Perceptron has an activation function (g) and a bias (mu). The value of a Perceptron is given by g(sum_i w_i X_i + mu), where X_i are the values of nodes input to this Perceptron with weights w_i.

Network training will use existing bias values for the starting values for the trainer. Upon completion of network training, the bias values are set to the values computed by the trainer.

See Also:
Feed Forward Class Example 1, Serialized Form

Method Summary
 Activation getActivation()
          Returns the activation function.
 double getBias()
          Returns the bias for this Perceptron.
 void setActivation(Activation activation)
          Sets the activation function.
 void setBias(double bias)
          Sets the bias for this Perceptron.
 
Methods inherited from class com.imsl.datamining.neural.Node
getLayer
 
Methods inherited from class java.lang.Object
clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait
 

Method Detail

getActivation

public Activation getActivation()
Returns the activation function.

Returns:
An Activation object indicating the activation function.

getBias

public double getBias()
Returns the bias for this Perceptron.

Returns:
A double representing the bias for this Perceptron.

setActivation

public void setActivation(Activation activation)
Sets the activation function.

Parameters:
activation - An Activation object which represents the activation g to be used by this Perceptron.

setBias

public void setBias(double bias)
Sets the bias for this Perceptron.

Parameters:
bias - A double scalar value to which the bias is to be set. The bias has a default value of 0.

JMSLTM Numerical Library 6.1

Copyright © 1970-2010 Visual Numerics, Inc.
Built July 30 2010.