5.2. FRAMEWORK 81
5.2.1 GVGAI LEARNING FRAMEWORK
Learning Phase N
L
N
Test Phase N
T
N
T
D N N
L
Learning phase. T
L
M
N
L
N
L
abort
result
abort
T
L
Test phase. N
T
5.2.2 GVGAI LEARNING ENVIRONMENT
Java Python
String PNG
lastSsoType = Types.LEARNING_SSO_TYPE.JSON; //request for a JSON
lastSsoType = Types.LEARNING_SSO_TYPE.IMAGE; //request for a
https://github.com/GAIGResearch/GVGAI
82 5. LEARNING IN GVGAI
screen-shot
lastSsoType = Types.LEARNING_SSO_TYPE.BOTH; //request for both
SerializableStateObservation phase=ACT, isValidation=false,
gameScore=8.0, gameTick=150, gameWinner=NO_WINNER, isGameOver=false,
worldDimension=[230.0, 200.0], blockSize=10, noOfPlayers=1,
...
availableActions=[ACTION_USE, ACTION_LEFT, ACTION_RIGHT],
avatarResources={}, observationGrid={ Observation{category=6, itype=0,
obsID=578, position=0.0 : 0.0, reference=-1.0 : -1.0, sqDist=2.0}
...
Observation{category=6, itype=11, obsID=733, position=150.0 : 40.0,
reference=-1.0 : -1.0, sqDist=24482.0} }, resourcesPositions=null,
portalsPositions=null, fromAvatarSpritesPositions=null}
Java Python
AbstractPlayer act init result
Agent.java Agent.py
..................Content has been hidden....................
You can't read the all page of ebook, please click here login for view all page.