burlap.behavior.singleagent.vfa.common

Class ConcatenatedObjectFeatureVectorGenerator

java.lang.Object

burlap.behavior.singleagent.vfa.common.ConcatenatedObjectFeatureVectorGenerator

All Implemented Interfaces:

StateToFeatureVectorGenerator

public class ConcatenatedObjectFeatureVectorGenerator
extends java.lang.Object
implements StateToFeatureVectorGenerator

This class is used to produce a state feature vector that is the concatenation of the observable attributes of objects belonging to a specified sequence of object classes. The feature vector attribtues are ordered first by the class of object and then by the order of the objects that appear for each object of that class. Values in the feature vector can optionally be specfied to be the normalized object values, which uses the attributes lowerlims and upperlims attribute to determine normalization.

Author:

James MacGlashan

Field Summary

Fields

Modifier and Type	Field and Description
protected boolean	normalizeValues Whether object values should be normalized.
protected java.lang.String[]	objectClassOrder The order of object classes and their order to follow when concatenting objects into a single state feature vector.

Constructor Summary

Constructors

Constructor and Description
ConcatenatedObjectFeatureVectorGenerator(boolean normalizeValues, java.lang.String... objectClassOrder) Initializes with an array of or object class names.
ConcatenatedObjectFeatureVectorGenerator(java.lang.String... objectClassOrder) Initializes with an array of or object class names.

Method Summary

Methods

Modifier and Type	Method and Description
double[]	generateFeatureVectorFrom(State s) Returns a feature vector represented as a double array for a given input state.
void	setNormalizeValues(boolean normalizeValues) Sets whether the object values are normalized in the returned feature vector

Methods inherited from class java.lang.Object

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

Field Detail

objectClassOrder

protected java.lang.String[] objectClassOrder

The order of object classes and their order to follow when concatenting objects into a single state feature vector.

normalizeValues

protected boolean normalizeValues

Whether object values should be normalized.

Constructor Detail

ConcatenatedObjectFeatureVectorGenerator

public ConcatenatedObjectFeatureVectorGenerator(java.lang.String... objectClassOrder)

Initializes with an array of or object class names. The resulting state feature vector will only be made up of objects beloning to the classes that appear in the array. Furthermore, the order that the observable values of the objects are concatenated will follow the order that these object classes are specfied, and then by the order that objects appear in the BURLAP state.

Parameters:

objectClassOrder - the sequence of object classes to use when constructing a state feature vector.

ConcatenatedObjectFeatureVectorGenerator

public ConcatenatedObjectFeatureVectorGenerator(boolean normalizeValues,
                                        java.lang.String... objectClassOrder)

Initializes with an array of or object class names. The resulting state feature vector will only be made up of objects beloning to the classes that appear in the array. Furthermore, the order that the observable values of the objects are concatenated will follow the order that these object classes are specfied, and then by the order that objects appear in the BURLAP state. If normalizeValues is set to true, then the value of all attributes are first normalized using the attribute lowerlims and upperlims fields.

Parameters:

normalizeValues - whether the values in the feature vector should be normalized.

objectClassOrder - the sequence of object classes to use when constructing a state feature vector.

Method Detail

setNormalizeValues

public void setNormalizeValues(boolean normalizeValues)

Sets whether the object values are normalized in the returned feature vector

Parameters:

normalizeValues - true if values should be normalized; false if they should be unnormalized.

generateFeatureVectorFrom

public double[] generateFeatureVectorFrom(State s)

Description copied from interface: StateToFeatureVectorGenerator

Returns a feature vector represented as a double array for a given input state.

Specified by:

generateFeatureVectorFrom in interface StateToFeatureVectorGenerator

Parameters:

s - the input state to turn into a feature vector.

Returns:

the feature vector represented as a double array.