A rough-set based incremental approach for updating attribute reduction under dynamic incomplete decision systems

Abstract

Efficient attribute reduction in large-scale incomplete decision systems is a challenging problem. The computation of tolerance classes induced by the condition attributes in the incomplete decision system is a key part among all existing attribute reduction algorithms. Moreover, updating attribute reduction for dynamically-increasing decision systems has attracted much attention, in view of that incremental attribute reduction algorithms in a dynamic incomplete decision system have not yet been sufficiently discussed so far. In this paper, we first introduce a simpler way of computing tolerance classes than the classical method. Then we present an incremental attribute reduction algorithm to compute an attribute reduct for a dynamically-increasing incomplete decision system. Compared with the non-incremental algorithms, our incremental attribute reduction algorithm can compute a new attribute reduct in much shorter time. Experiments on four data sets downloaded from UCI show that the feasibility and effectiveness of the proposed incremental algorithm.