Inclusive Pruning: a New Class of Pruning Axiom for Unordered Search and Its Application to Classification Learning

This paper presents a new class of pruning axiom for unordered search. Previous pruning axioms for unordered search identify operators that should not be applied in order to prune states reached via those operators. In contrast, the new pruning axioms identify operators that should be applied and prune states that are not reached via those operators. Specific pruning axioms employing both these approaches are identified for classification learning. Experimental results demonstrate that application of the new pruning axioms can reduce by more than 60% the number of states from the search space that are considered during classification learning.