org.encog.neural.pattern
Class BAMPattern

java.lang.Object
  extended by org.encog.neural.pattern.BAMPattern
All Implemented Interfaces:
NeuralNetworkPattern

public class BAMPattern
extends Object
implements NeuralNetworkPattern

Construct a Bidirectional Access Memory (BAM) neural network. This neural network type learns to associate one pattern with another. The two patterns do not need to be of the same length. This network has two that are connected to each other. Though they are labeled as input and output layers to Encog, they are both equal, and should simply be thought of as the two layers that make up the net.


Constructor Summary
BAMPattern()
           
 
Method Summary
 void addHiddenLayer(int count)
          Unused, a BAM has no hidden layers.
 void clear()
          Clear any settings on the pattern.
 MLMethod generate()
          Generate the specified neural network.
 void setActivationFunction(ActivationFunction activation)
          Not used, the BAM uses a bipoloar activation function.
 void setF1Neurons(int count)
          Set the F1 neurons.
 void setF2Neurons(int count)
          Set the output neurons.
 void setInputNeurons(int count)
          Set the number of input neurons.
 void setOutputNeurons(int count)
          Set the number of output neurons.
 
Methods inherited from class java.lang.Object
clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait
 

Constructor Detail

BAMPattern

public BAMPattern()
Method Detail

addHiddenLayer

public void addHiddenLayer(int count)
Unused, a BAM has no hidden layers.

Specified by:
addHiddenLayer in interface NeuralNetworkPattern
Parameters:
count - Not used.

clear

public void clear()
Clear any settings on the pattern.

Specified by:
clear in interface NeuralNetworkPattern

generate

public MLMethod generate()
Description copied from interface: NeuralNetworkPattern
Generate the specified neural network.

Specified by:
generate in interface NeuralNetworkPattern
Returns:
The generated network.

setActivationFunction

public void setActivationFunction(ActivationFunction activation)
Not used, the BAM uses a bipoloar activation function.

Specified by:
setActivationFunction in interface NeuralNetworkPattern
Parameters:
activation - Not used.

setF1Neurons

public void setF1Neurons(int count)
Set the F1 neurons. The BAM really does not have an input and output layer, so this is simply setting the number of neurons that are in the first layer.

Parameters:
count - The number of neurons in the first layer.

setF2Neurons

public void setF2Neurons(int count)
Set the output neurons. The BAM really does not have an input and output layer, so this is simply setting the number of neurons that are in the second layer.

Parameters:
count - The number of neurons in the second layer.

setInputNeurons

public void setInputNeurons(int count)
Set the number of input neurons.

Specified by:
setInputNeurons in interface NeuralNetworkPattern
Parameters:
count - The number of input neurons.

setOutputNeurons

public void setOutputNeurons(int count)
Set the number of output neurons.

Specified by:
setOutputNeurons in interface NeuralNetworkPattern
Parameters:
count - The output neuron count.


Copyright © 2014. All Rights Reserved.