APPROXIMATIONS BASED MACHINE LEARNING APPROACHES IN INCOMPLETE VAGUE DECISION TABLE

The vagueness and indiscernibility constitute two aspects of uncertainty in decision table.With a rapid growth of interest in recent years, vague set theory has become an effective tool to handle inexact data in the fields of fuzzy information processing. However, machine learning approaches aiming at vague sets have not been reported yet. Most of the existing researches on vague sets still focus on discussing their properties. In this paper, some basic notions of vague sets are introduced. And then benefiting from the idea of approximations in rough set theory, a learning mechanism of vague decision table is presented, especially for the incomplete table with unknown values. It has solved the learning problems of inexact data. The proposed algorithms are suitable for decision attribute values with precise and vague data respectively.