org.aika.training

Class LongTermLearning

java.lang.Object

org.aika.training.LongTermLearning

public class LongTermLearning
extends Object

Implements the biologically inspired learning algorithms for long-term potentiation and long-term depression. Wikipedia: "In neuroscience, long-term potentiation (LTP) is a persistent strengthening of synapses based on recent patterns of activity. These are patterns of synaptic activity that produce a long-lasting increase in signal transmission between two neurons. The opposite of LTP is long-term depression, which produces a long-lasting decrease in synaptic strength."

Author:

Lukas Molzberger

Nested Class Summary

Nested Classes

Modifier and Type	Class and Description
static class	LongTermLearning.Config

Constructor Summary

Constructors

Constructor and Description
LongTermLearning()

Method Summary

Modifier and Type	Method and Description
static void	longTermDepression(Document doc, LongTermLearning.Config config, Activation act, boolean dir) The long-term depression algorithm decreases the strength of a synapse if only one side of the synapse is firing.
static void	longTermPotentiation(Document doc, LongTermLearning.Config config, Activation act) The long-term potentiation algorithm is a variant of the Hebb learning rule.
static void	train(Document doc, LongTermLearning.Config config)

Methods inherited from class java.lang.Object

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

Constructor Detail

LongTermLearning

public LongTermLearning()

Method Detail

train

public static void train(Document doc,
                         LongTermLearning.Config config)

longTermPotentiation

public static void longTermPotentiation(Document doc,
                                        LongTermLearning.Config config,
                                        Activation act)

The long-term potentiation algorithm is a variant of the Hebb learning rule.

Parameters:

doc -

config -

act -

longTermDepression

public static void longTermDepression(Document doc,
                                      LongTermLearning.Config config,
                                      Activation act,
                                      boolean dir)

The long-term depression algorithm decreases the strength of a synapse if only one side of the synapse is firing. The algorithm tries however to preserve the logical characteristic of the synapse. If for example the synapse has an or-characteristic, then a non firing input neuron and a firing output neuron will not change the synapse weight. On the other hand, if the synapse has an and-characteristic, then a firing input neuron and a non firing output neuron will not change the synapse weight, too.

Parameters:

doc -

config -

act -

dir -