A rule generation algorithm based on granular computing
Jiu-Jiang An, Guoyin Wang, Yu Chuan Wu, Quan Gan · 2005
Granular computing has been applied in many fields to solve problems or describe problem spaces at different granularities and hierarchies. In this paper, a rule generation algorithm based on granular computing (RGAGC) is developed. RGAGC is a valid method to generate rules from the granule space. Compared with many classic decision tree algorithms, RGAGC generates a single rule granule in each step instead of selecting a suitable attribute. It is a more general algorithm for rule generation, since it could generate rules from the granule space without considering the problem of selecting an attribute according to some measure. On the other hand, in order to improve the performance of rule granule generation, the "false preserving" property of quotient space theory is used as a strategy to control the process of rule granule generation, so that RGAGC could generate rule granules from the granule space quickly. Our simulation experiment results prove that RGAGC is valid.