org.dllearner.algorithms.refinement2
Class FlexibleHeuristic

java.lang.Object
  org.dllearner.algorithms.refinement2.FlexibleHeuristic

All Implemented Interfaces:

Comparator<ExampleBasedNode>, ExampleBasedHeuristic

public class FlexibleHeuristic
extends Object
implements ExampleBasedHeuristic

This heuristic compares two nodes by computing a score using the number of covered negatives and the horizontal expansion factor of a node as input. Using this score it decides which one of the nodes seems to be more promising. The heuristic is flexible, because it offers a tradeoff between accurary and horizontal expansion (concept length). In contrast to the lexicographic heuristic this means that it sometimes prefers worse classifiers with low horizontal expansion over a better classifier with high horizontal expansion. It can be configured by using the "percentPerLenghtUnit" constructor argument. A higher value means that the algorithm is more likely to search in unexplored areas (= low horizontal expansion) of the search space vs. looking in promising but already explored (= high horizontal expansion) areas of the search space.

Author:

Jens Lehmann

Constructor Summary

FlexibleHeuristic(int nrOfNegativeExamples, double percentPerLengthUnit)

Method Summary

int	compare(ExampleBasedNode n1, ExampleBasedNode n2)
boolean	equals(Object o)

Methods inherited from class java.lang.Object

getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Constructor Detail

FlexibleHeuristic

public FlexibleHeuristic(int nrOfNegativeExamples,
                         double percentPerLengthUnit)

Method Detail

compare

public int compare(ExampleBasedNode n1,
                   ExampleBasedNode n2)

Specified by:

compare in interface Comparator<ExampleBasedNode>

equals

public boolean equals(Object o)

Specified by:

equals in interface Comparator<ExampleBasedNode>

Overrides:

equals in class Object