Incremental Computing Approximations with the Dynamic Object set in Interval-valued Ordered Information System

Abstract

Rough set theory has been successfully used in formation system for classification analysis and knowledge discovery. The upper and lower approximations are fundamental concepts of this theory. The new information arrives continuously and redundant information may be produced with the time in real-world application. So, then incremental learning is an efficient technique for knowledge discovery in a dynamic database, which enables acquiring additional knowledge from new data without forgetting prior knowledge, which need to be updated incrementally while the object set get varies over time in the interval-valued ordered information system. In this paper, we analyzed the updating mechanisms for computing approximations with the variation of the object set. Two incremental algorithms respectively for adding and deleting objects with updating the approximations are proposed in interval-valued ordered information system. Furthermore, extensive experiments are carried out on six UCI data sets to verify the performance of these proposed algorithms. And the experiments results indicate the incremental approaches significantly outperform non-incremental approaches with a dramatic reduction in the computational time.