burlap.behavior.singleagent.vfa.common

Class FVToFeatureDatabase

java.lang.Object

burlap.behavior.singleagent.vfa.common.FVToFeatureDatabase

All Implemented Interfaces:

FeatureDatabase

public class FVToFeatureDatabase
extends java.lang.Object
implements FeatureDatabase

A class wrapper for converting a StateToFeatureVectorGenerator into a FeatureDatabase. One advantage of this approach is that this can be used to automatically construct state-action features. Specifically, the state features produced but the StateToFeatureVectorGenerator are duplicated into unique features for each possible action. The duplication of features for actions is lazy. That is, every time a previously unseen action is queried in the getActionFeaturesSets(burlap.oomdp.core.states.State, java.util.List) method, it gets new state features assigned to it.

Author:

James MacGlashan.

Field Summary

Fields

Modifier and Type	Field and Description
protected java.util.Map<GroundedAction,java.lang.Integer>	actionFeatureMultiplier A map for returning a multiplier to the number of state features for each action.
protected int	dim The dimensionality of the state features produced by the feature vector generator
protected StateToFeatureVectorGenerator	fvGen The state feature vector generator
protected int	nextActionMultiplier The next action Fourier basis function size multiplier to use for the next newly seen action.

Constructor Summary

Constructors

Constructor and Description
FVToFeatureDatabase(StateToFeatureVectorGenerator fvGen, int dim) Initializes.

Method Summary

Methods

Modifier and Type	Method and Description
FVToFeatureDatabase	copy() Returns a deep copy of this feature database.
void	freezeDatabaseState(boolean toggle) Enable or disable the database from generating new features.
java.util.List<ActionFeaturesQuery>	getActionFeaturesSets(State s, java.util.List<GroundedAction> actions) Returns non-zero action features for all of provided actions in state s.
protected int	getActionMultiplier(GroundedAction ga) This method returns the action multiplier for the specified grounded action.
java.util.List<StateFeature>	getStateFeatures(State s) Returns non-zero state features for a given state.
int	numberOfFeatures() Returns the number of features this database tracks.

Methods inherited from class java.lang.Object

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

Field Detail

fvGen

protected StateToFeatureVectorGenerator fvGen

The state feature vector generator

dim

protected int dim

The dimensionality of the state features produced by the feature vector generator

actionFeatureMultiplier

protected java.util.Map<GroundedAction,java.lang.Integer> actionFeatureMultiplier

A map for returning a multiplier to the number of state features for each action. Effectively this ensures a unique feature ID for each state feature for each action.

nextActionMultiplier

protected int nextActionMultiplier

The next action Fourier basis function size multiplier to use for the next newly seen action.

Constructor Detail

FVToFeatureDatabase

public FVToFeatureDatabase(StateToFeatureVectorGenerator fvGen,
                   int dim)

Initializes.

Parameters:

fvGen - The state feature vector generator

dim - The dimensionality of the state features produced by the feature vector generator

Method Detail

getStateFeatures

public java.util.List<StateFeature> getStateFeatures(State s)

Description copied from interface: FeatureDatabase

Returns non-zero state features for a given state. This method should be implemented if it is to be used with algorithms that compute V-values from state features or other values that are independent of the actions.

Specified by:

getStateFeatures in interface FeatureDatabase

Parameters:

s - the state for which features should be returned

Returns:

the features for state s

getActionFeaturesSets

public java.util.List<ActionFeaturesQuery> getActionFeaturesSets(State s,
                                                        java.util.List<GroundedAction> actions)

Description copied from interface: FeatureDatabase

Returns non-zero action features for all of provided actions in state s. This method should be implemented if it is to be used with algorithms that compute Q-values from action features or other values are the dependent on the actions. Note that features for different actions should return different feature ids.

Specified by:

getActionFeaturesSets in interface FeatureDatabase

Parameters:

s - the state for which features should be returned

actions - the action set for which the action features should be returned

Returns:

returns the set set of action features for each action in actions

freezeDatabaseState

public void freezeDatabaseState(boolean toggle)

Description copied from interface: FeatureDatabase

Enable or disable the database from generating new features. This method only needs to do anything if it creates features in an online fashion. That is, if the set of possible state features is defined at construction, then this method does not need to do anything. If features are generated as needed, such as with an instance-based feature set, then when the database is told to be frozen no new instances should be created.

Specified by:

freezeDatabaseState in interface FeatureDatabase

numberOfFeatures

public int numberOfFeatures()

Description copied from interface: FeatureDatabase

Returns the number of features this database tracks.

Specified by:

numberOfFeatures in interface FeatureDatabase

Returns:

the number of features this database tracks.

getActionMultiplier

protected int getActionMultiplier(GroundedAction ga)

This method returns the action multiplier for the specified grounded action. If the action is not stored, a new action multiplier will created, stored, and returned.

Parameters:

ga - the grounded action for which the multiplier will be returned

Returns:

the action multiplier to be applied to a state feature id.

copy

public FVToFeatureDatabase copy()

Description copied from interface: FeatureDatabase

Returns a deep copy of this feature database. If the feature database is dynamic, then changes to one will not affect the other.

Specified by:

copy in interface FeatureDatabase

Returns:

a deep copy of this feature database