org.aika.lattice

Class Node

java.lang.Object

org.aika.lattice.Node

All Implemented Interfaces:

Comparable<Node>, Writable

Direct Known Subclasses:

AndNode, InputNode, OrNode

public abstract class Node
extends Object
implements Comparable<Node>, Writable

The Node class is the abstract class for all the boolean logic nodes underneath the neural network layer. These nodes form a boolean representation for all the neurons of the neural network. Whenever changes occur to the synapse weights in the neural layer, then the structure of the boolean representation needs to be adjusted. Several neurons, however, might share common substructures in this boolean representation. The InputNode and the AndNode classes together form a pattern lattice, containing all possible substructures of any given conjunction. For example if we have the conjunction ABCD where A, B, C, D are the inputs then the pattern lattice will contain the nodes ABCD, ABC, ABD, ACD, BCD, AB, AC, AD, BC, BD, CD, A, B, C, D. The class OrNode is a disjunction of either input-nodes or and-nodes. The or-node is connected with one of the neurons. Each logic node has a set of activations. The activations are stored in the thread local data structure ThreadState.

Author:

Lukas Molzberger

Nested Class Summary

Nested Classes

Modifier and Type	Class and Description
static class	Node.ReverseAndRefinement
static class	Node.RidVisited
static class	Node.RSKey
static class	Node.Similarity
static class	Node.ThreadState

Field Summary

Fields

Modifier and Type	Field and Description
TreeMap<AndNode.Refinement,AndNode>	andChildren
static Comparator<Activation.Key>	BEGIN_COMP
static AtomicInteger	currentNodeId
static Comparator<Activation.Key>	END_COMP
boolean	endRequired
int	frequency
boolean	frequencyHasChanged
int	id
int	instanceSum
boolean	isBlocked
boolean	isQueued
boolean	isRemoved
int	isRemovedId
static int	isRemovedIdCounter
int	level
static boolean	LINK_NEURON_RELATIONS_OPTIMIZATION
ReadWriteLock	lock
static org.aika.lattice.Node.DummyNode	MAX_NODE
static int	MAX_RID
static org.aika.lattice.Node.DummyNode	MIN_NODE
static int	minFrequency
Neuron	neuron
int	nOffset
double	nullHypFreq
int	numberOfNeuronRefs
double	oldNullHypFreq
TreeSet<OrNode.OrEntry>	orChildren
boolean	passive
long	queueId
TreeMap<Node.ReverseAndRefinement,AndNode.Refinement>	reverseAndChildren
static Comparator<Activation.Key>	RID_COMP
boolean	ridRequired
int	sizeSum
Node.ThreadState[]	threads
static long	visitedCounter

Constructor Summary

Constructors

Modifier	Constructor and Description
protected	Node()
	Node(Document doc, int level)

Method Summary

Modifier and Type	Method and Description
static void	addActivationAndPropagate(Document doc, Activation.Key ak, Collection<Activation> inputActs)
Activation	addActivationInternal(Document doc, Activation.Key ak, Collection<Activation> inputActs, boolean isTrainingAct)
void	addAndChild(AndNode.Refinement ref, AndNode child)
static void	addConflict(Document doc, InterprNode io, InterprNode o, Activation act, Collection<Activation> inputActs, long v)
void	addOrChild(Document doc, OrNode.OrEntry oe)
static boolean	adjust(Document doc, Neuron neuron, int dir)
abstract void	apply(Document doc, Activation act, InterprNode conflict)
protected abstract void	changeNumberOfNeuronRefs(Document doc, long v, int d)
abstract void	cleanup(Document doc)
void	clearActivations(Document doc)
void	clearActivations(Model m)
protected abstract Collection<AndNode.Refinement>	collectNodeAndRefinements(AndNode.Refinement newRef)
static int	compare(Node a, Node b)
int	compareTo(Node n)
boolean	computeAndParents(Document doc, Integer offset, SortedSet<AndNode.Refinement> inputs, Map<AndNode.Refinement,Node> parents, boolean discoverPatterns, long v)
abstract void	computeNullHyp(Model m)
static void	computeRefinements(Document doc, TreeSet<Node.RSKey> queue, Neuron n, Node.RSKey rsk, long v, List<Node.RSKey> outputs, List<Node.RSKey> cleanup)
abstract double	computeSynapseWeightSum(Integer offset, Neuron n)
void	count(Document doc)
abstract void	deleteActivation(Document doc, Activation act)
abstract void	discover(Document doc, Activation act)
Collection<Activation>	getActivations(Document doc)
AndNode	getAndChild(AndNode.Refinement ref)
Activation	getFirstActivation(Document doc)
Node.ThreadState	getThreadState(Document doc, boolean create)
protected abstract boolean	hasSupport(Activation act)
abstract void	initActivation(Document doc, Activation act)
abstract boolean	isAllowedOption(Document doc, InterprNode n, Activation act, long v)
boolean	isCovered(Document doc, Integer offset, long v)
abstract boolean	isExpandable(boolean checkFrequency)
boolean	isFrequent()
boolean	isPublic()
boolean	isRequired()
abstract String	logicToString()
protected void	postProcessRemovedActivation(Document doc, Activation act, Collection<Activation> inputActs)
protected void	prepareResultsForPredefinedNodes(Document doc, TreeSet<Node.RSKey> queue, long v, List<Node.RSKey> outputs, List<Node.RSKey> cleanup, Neuron n, Synapse s, Integer offset)
protected Range	preProcessAddedActivation(Document doc, Activation.Key ak, Collection<Activation> inputActs)
void	processAddedActivation(Document doc, Activation.Key ak, Collection<Activation> inputActs)
void	processChanges(Document doc)
void	processRemovedActivation(Document doc, Activation act, Collection<Activation> inputActs)
abstract void	propagateAddedActivation(Document doc, Activation act, InterprNode conflict)
abstract void	propagateRemovedActivation(Document doc, Activation act)
static Node	read(DataInput in, Document doc)
void	readFields(DataInput in, Document doc) Deserialize the fields of this object from in.
void	remove(Document doc)
static void	removeActivationAndPropagate(Document doc, Activation act, Collection<Activation> inputActs)
boolean	removeActivationInternal(Document doc, Activation act, Collection<Activation> inputActs)
void	removeAndChild(AndNode.Refinement ref)
static void	removeConflict(Document doc, InterprNode io, InterprNode o, Activation act, Activation nAct, long v)
void	removeFromNextLevel(Document doc, Activation iAct)
void	removeOrChild(Document doc, OrNode.OrEntry oe)
String	toSimpleString()
String	toString()
String	weightsToString()
void	write(DataOutput out) Serialize the fields of this object to out.

Methods inherited from class java.lang.Object

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

Field Detail

minFrequency

public static int minFrequency

MAX_RID

public static int MAX_RID

LINK_NEURON_RELATIONS_OPTIMIZATION

public static boolean LINK_NEURON_RELATIONS_OPTIMIZATION

MIN_NODE

public static final org.aika.lattice.Node.DummyNode MIN_NODE

MAX_NODE

public static final org.aika.lattice.Node.DummyNode MAX_NODE

currentNodeId

public static AtomicInteger currentNodeId

id

public int id

reverseAndChildren

public TreeMap<Node.ReverseAndRefinement,AndNode.Refinement> reverseAndChildren

andChildren

public TreeMap<AndNode.Refinement,AndNode> andChildren

orChildren

public TreeSet<OrNode.OrEntry> orChildren

level

public int level

passive

public boolean passive

frequency

public volatile int frequency

nullHypFreq

public volatile double nullHypFreq

oldNullHypFreq

public volatile double oldNullHypFreq

isBlocked

public boolean isBlocked

endRequired

public boolean endRequired

ridRequired

public boolean ridRequired

numberOfNeuronRefs

public int numberOfNeuronRefs

isRemoved

public volatile boolean isRemoved

isRemovedId

public volatile int isRemovedId

isRemovedIdCounter

public static int isRemovedIdCounter

frequencyHasChanged

public volatile boolean frequencyHasChanged

nOffset

public volatile int nOffset

sizeSum

public volatile int sizeSum

instanceSum

public volatile int instanceSum

lock

public ReadWriteLock lock

isQueued

public boolean isQueued

queueId

public long queueId

neuron

public Neuron neuron

visitedCounter

public static long visitedCounter

threads

public Node.ThreadState[] threads

BEGIN_COMP

public static final Comparator<Activation.Key> BEGIN_COMP

END_COMP

public static final Comparator<Activation.Key> END_COMP

RID_COMP

public static final Comparator<Activation.Key> RID_COMP

Constructor Detail

Node

protected Node()

Node

public Node(Document doc,
            int level)

Method Detail

getThreadState

public Node.ThreadState getThreadState(Document doc,
                                       boolean create)

propagateAddedActivation

public abstract void propagateAddedActivation(Document doc,
                                              Activation act,
                                              InterprNode conflict)

propagateRemovedActivation

public abstract void propagateRemovedActivation(Document doc,
                                                Activation act)

isAllowedOption

public abstract boolean isAllowedOption(Document doc,
                                        InterprNode n,
                                        Activation act,
                                        long v)

cleanup

public abstract void cleanup(Document doc)

initActivation

public abstract void initActivation(Document doc,
                                    Activation act)

deleteActivation

public abstract void deleteActivation(Document doc,
                                      Activation act)

computeSynapseWeightSum

public abstract double computeSynapseWeightSum(Integer offset,
                                               Neuron n)

logicToString

public abstract String logicToString()

apply

public abstract void apply(Document doc,
                           Activation act,
                           InterprNode conflict)

discover

public abstract void discover(Document doc,
                              Activation act)

collectNodeAndRefinements

protected abstract Collection<AndNode.Refinement> collectNodeAndRefinements(AndNode.Refinement newRef)

changeNumberOfNeuronRefs

protected abstract void changeNumberOfNeuronRefs(Document doc,
                                                 long v,
                                                 int d)

hasSupport

protected abstract boolean hasSupport(Activation act)

computeNullHyp

public abstract void computeNullHyp(Model m)

isExpandable

public abstract boolean isExpandable(boolean checkFrequency)

addOrChild

public void addOrChild(Document doc,
                       OrNode.OrEntry oe)

removeOrChild

public void removeOrChild(Document doc,
                          OrNode.OrEntry oe)

addAndChild

public void addAndChild(AndNode.Refinement ref,
                        AndNode child)

removeAndChild

public void removeAndChild(AndNode.Refinement ref)

count

public void count(Document doc)

addActivationInternal

public Activation addActivationInternal(Document doc,
                                        Activation.Key ak,
                                        Collection<Activation> inputActs,
                                        boolean isTrainingAct)

removeActivationInternal

public boolean removeActivationInternal(Document doc,
                                        Activation act,
                                        Collection<Activation> inputActs)

addConflict

public static void addConflict(Document doc,
                               InterprNode io,
                               InterprNode o,
                               Activation act,
                               Collection<Activation> inputActs,
                               long v)

removeConflict

public static void removeConflict(Document doc,
                                  InterprNode io,
                                  InterprNode o,
                                  Activation act,
                                  Activation nAct,
                                  long v)

processChanges

public void processChanges(Document doc)

addActivationAndPropagate

public static void addActivationAndPropagate(Document doc,
                                             Activation.Key ak,
                                             Collection<Activation> inputActs)

preProcessAddedActivation

protected Range preProcessAddedActivation(Document doc,
                                          Activation.Key ak,
                                          Collection<Activation> inputActs)

processAddedActivation

public void processAddedActivation(Document doc,
                                   Activation.Key ak,
                                   Collection<Activation> inputActs)

removeActivationAndPropagate

public static void removeActivationAndPropagate(Document doc,
                                                Activation act,
                                                Collection<Activation> inputActs)

postProcessRemovedActivation

protected void postProcessRemovedActivation(Document doc,
                                            Activation act,
                                            Collection<Activation> inputActs)

processRemovedActivation

public void processRemovedActivation(Document doc,
                                     Activation act,
                                     Collection<Activation> inputActs)

getActivations

public Collection<Activation> getActivations(Document doc)

getFirstActivation

public Activation getFirstActivation(Document doc)

clearActivations

public void clearActivations(Document doc)

clearActivations

public void clearActivations(Model m)

isFrequent

public boolean isFrequent()

isPublic

public boolean isPublic()

computeAndParents

public boolean computeAndParents(Document doc,
                                 Integer offset,
                                 SortedSet<AndNode.Refinement> inputs,
                                 Map<AndNode.Refinement,Node> parents,
                                 boolean discoverPatterns,
                                 long v)

removeFromNextLevel

public void removeFromNextLevel(Document doc,
                                Activation iAct)

remove

public void remove(Document doc)

getAndChild

public AndNode getAndChild(AndNode.Refinement ref)

adjust

public static boolean adjust(Document doc,
                             Neuron neuron,
                             int dir)

computeRefinements

public static void computeRefinements(Document doc,
                                      TreeSet<Node.RSKey> queue,
                                      Neuron n,
                                      Node.RSKey rsk,
                                      long v,
                                      List<Node.RSKey> outputs,
                                      List<Node.RSKey> cleanup)

prepareResultsForPredefinedNodes

protected void prepareResultsForPredefinedNodes(Document doc,
                                                TreeSet<Node.RSKey> queue,
                                                long v,
                                                List<Node.RSKey> outputs,
                                                List<Node.RSKey> cleanup,
                                                Neuron n,
                                                Synapse s,
                                                Integer offset)

isCovered

public boolean isCovered(Document doc,
                         Integer offset,
                         long v)

isRequired

public boolean isRequired()

toString

public String toString()

Overrides:

toString in class Object

toSimpleString

public String toSimpleString()

weightsToString

public String weightsToString()

write

public void write(DataOutput out)
           throws IOException

Description copied from interface: Writable

Serialize the fields of this object to out.

Specified by:

write in interface Writable

Parameters:

out - DataOuput to serialize this object into.

Throws:

IOException

readFields

public void readFields(DataInput in,
                       Document doc)
                throws IOException

Description copied from interface: Writable

Deserialize the fields of this object from in.

For efficiency, implementations should attempt to re-use storage in the existing object where possible.

Specified by:

readFields in interface Writable

Parameters:

in - DataInput to deseriablize this object from.

Throws:

IOException

read

public static Node read(DataInput in,
                        Document doc)
                 throws IOException

Throws:

IOException

compareTo

public int compareTo(Node n)

Specified by:

compareTo in interface Comparable<Node>

compare

public static int compare(Node a,
                          Node b)