|
|||||||||||
PREV NEXT | FRAMES NO FRAMES |
AnjiNet
ANN.AnjiNet
to a FloatingEye
.
AnjiNet
) from
a chromosome.activator
with stimuli, and return results
a_chromosomeToAdd
to set of chromosomes to be evaluated.
elite
list.
c
.
input
with no transformation.
contestants
and a destination for losers.
a_activeConfiguration
.
init()
should be called before using this object
responses
and target values.
responses
from targets, sum all differences, subtract from max
fitness, and square result.
newMove
is empty, false otherwise.
addContestant()
com.anji.fingerprint
contains classes to handle fingerprint classification.com.anji.floatingeye
contains classes to build and operate a
"floating eye" which is an agent that can move up and down, left and right, and zoom
in and out relative to an image, or "surface".com.anji.gomoku
contains classes to handle Go-Moku (see Gamerz.net for rules).com.anji.imaging
contains utility classes to handle image processing.com.anji.integration
contains classes that provide the glue between the various components of the system, including artificial neural net implementations, JGAP, NEAT, Log4J, and persistence and presentation layers.com.anji.neat
contains classes implementing NEAT framework, including genes, mutation operators, and fitness function used for XOR.com.anji.nn
contains classes implementing a simple artificial neural net framework.com.anji.persistence
contains classes implementing a simple persistence interface.com.anji.polebalance
contains classes to implement pole balancing.com.anji.roshambo
contains classes to implement the game of
roshambo, also commonly known as
rock-scissors-paper.com.anji.run
contains classes concerned with an evolutionary run.com.anji.tournament
contains classes to handle tournaments and games.com.anji.ttt
contains classes to handle Tic-Tac-Toe (see Yahooligans for rules).com.anji.util
contains utility classes used in the ANJI framework.ConnectionGene
object
responses
and target values.
keepers
init()
init()
should be called before using this object
aMatch
countOpponentResults to
true
id
target
according to NEAT speciation
methodology.
target
according to NEAT speciation
methodology.
init()
before using this object
MAX_FITNESS
relative to # images ANN (transcribed
from c
) is able to identify correctly.
MAX_FITNESS
inclusive.
MAX_FITNESS
relative to # images ANNs (transcribed
from ensemble
) are able to identify correctly.
BulkFitnessFunction
insteadAnjiNet
ANN.type
and key
.
init()
to initialize
Genotype
and XML.aSize
init()
after ctor
init()
after ctor
idx
.
genes
as List
alleles
as SortedMap
contestants.size()
and double each round
genes
as List
genes
; otherwise, returns only
neuron genes of type
alleles
as SortedMap
alleles
; otherwise, returns only
neurons of type
addContestant()
thus far and associated results.
prefix
.
init()
before using this object
Player
that determines moves by input from stdin.aFileName
InvalidConfigurationException
, but in runtime form so it can be thrown from
methods that do not throw exceptions.init()
after this
newResults
.
newLosses
to losses
newRawScore
to rawScore
newTies
to ties
newWins
to wins
aMatch
countOpponentResults to true
aMatch
countOpponentResults to true
FloatingEye
.init()
aMatch
countOpponentResults to false
recurrent
property settings.
Logger
boardState
as input to activator and place token in empty space
corresponding to strongest output.
boardState
, with every possible next move filled in, as
input to activator and place token in empty space corresponding to strongest output.
boardState
rotated in each of the for cardinal directions, as
input to activator and place token in empty space corresponding to strongest output.
genesToAdd
and adds to genesToRemove
all
connection genes that are modified.
material
unmodified, but updates allelesToAdd
and
allelesToRemove
with modifications.
Chromosome
object (loaded from persistence if necessary) into
Activator
object and activate it with specified stimuli.aFunc
activation function.
anIncoming
and weight aWeight
.
NeuronGene
object
Activator
phenotype from Chromosome
genotype.
AnjiNet
from genotype
srcNeuronId
to
neuron destNeuronId
according to NEAT add connection mutation; if a previous
mutation has occurred adding a connection between srcNeuronId and destNeuronId, returns
connection with that id; otherwise, new innovation id
type
connectionId
according to NEAT add neuron mutation.
IllegalArgumentException
exception if key not present.
dimension
inputs.
aValues
as inputs.
PlayerResult
objects in descending order of score.values
.
Logger
contestants
.
srcNeuronId
and destNeuronId
and the connection that mutated between them via NEAT add connection mutation,
newConnectionId
connectionId
and the neuron that replaced
it via NEAT add neuron mutation, newNeuronId
Random
object to ensure all of system is using same random
sequence.init()
after ctor
Game
interface.init(Properties)
after this ctor
RoshamboPlayer
objects from AniNet
objects
transcribed from Chromosome
objects.init()
init()
after this ctor, unless it's called from hibernate
parents
to offspring
.
parents
to offspring
.
parents
to offspring
.
maxWeightRemoved
Player
contestants, has them play each
other in matches, and packages the results.init()
PlayerAndResults
for each game.init()
should be called before using this object
aMatch
countOpponentResults to
true
init()
after this constructor
a_howManyToSelect
chromosomes with highest speciated fitness.
BufferedImage
; this image is larger
than specified dimension so floating eye can go off the edge; these "off the edge" spaces
are set to aNonviewableSpaceValue
MIN_INIT_WEIGHT
and
MAX_INIT_WEIGHT
idx
to value
.
genotype
.
toString()
Activator
output is to a target.init(Properties)
after this
anActivator
, 10 inputs and 9 outputs.
anActivator
, 10 inputs and 1 output.
anActivator
, 10 inputs and 9 outputs.
coords
based on perspective; coords
should be
between 0 and max, inclusive
pos
is value
affinity should be between 0.0 and 1.0 inclusive
newMove
in array boardState
.
newMove
will now equal 1
boardState
at position newMove
.
material
with specified sets of alleles; alleles present in both lists
will be added (or replaced if the gene existed on original material)
TestNeuron
only
protected
visibility increases performance of
PatternConnection.read()
init()
before using this object
|
|||||||||||
PREV NEXT | FRAMES NO FRAMES |