Induction in an abstraction space: a form of constructive induction

摘要: We report on a learning system MIRO which performs supervised concept formation in an abstraction space. Given domain theory, the method constructs this space by deduction over instances, and then induction it rather than initial defined instances alone. It is also possible to regard as variant of constructive induction. The Vapnik-Chervonenkis model suggests that can result substantial speedup, we provide empirical studies validate proposition. show reduce number false negative postive classifications because coincidental patterns are filtered process. able extend incomplete theory represented at tribute-value pairs with set rules represent disjunctive derived from batch training instances.