Constructing Nominal X-of-N Attributes

Abstract

Most constructive induction researchers focus only on new boolean attributes. This paper reports a new constructive induction algorithm, called XofN, that constructs new nominal attributes in the form of X-of-N representations. An X-of-N is a set containing one or more attribute-value pairs. For a given instance, its value corresponds to the number of its attribute-value pairs that are true. The promising preliminary experimental results, on both artificial and real-world domains, show that constructing new nominal attributes in the form of X-of-N representations can significantly improve the performance of selective induction in terms of both higher prediction accuracy and lower theory complexity. 1 Introduction A well-known elementary limitation of selective induction algorithms is that when task-supplied attributes are not adequate for describing hypotheses, their performance in terms of prediction accuracy and/or theory complexity is poor. To overcome this limitation, constructiv...