MINIMAL AND MAXIMAL RULES LEARNING AND ITS APPLICATION TO SIMPLIFYING DECISION TREE RULES

The notions of minimal rule and maximal rule,especially the latter,are addressed in the light of the notion of reduction from rough set.And then a new learning algorithm called minimal and maximal rules learning is proposed.It can be used as a post processing to simplify the rules generated by decision tree induction,which currently is the most efficient classification method in KDD.The empirical result illustrates that the final decision tree rules simplified by the new method are considerably simpler than before.