A Minimization Approach to Propositional Inductive Learning

Abstract

An approach to the problem of propositional inductive learning of if-then-else rules, different from the commonly used ones, is presented in the paper. Main differences are: literal selection process that searches for the smallest set of literals so that the completely correct rule for all learning examples can be constructed and the nonnormal form of the generated rules built by the search for necessary and sufficient conditions of example classes. It is also presented how iterative application of the literal selection process can solve the problem of learning from noisy domains by appropriate exclusion of some learning examples. The results of application of the system that includes described algorithms on a few publicly available domains are discussed.