基于可变相容粒空间的多粒度覆盖粗糙集模型

Multi-granule Covering Rough Sets Model Based on Variable Precision Tolerant-granule Space

在研究分析了以往对覆盖粗糙集处理方法的基础上,针对目前研究只是限制在改进对象间的关系,优化寻找对象关系类的条件领域,而论域覆盖空间的限制条件太过宽泛这一问题,文中结合贝叶斯全概率的思想,充分考虑条件和结果存在的内在必然联系,提出一种基于量化容差关系的相容粒空间。将相容粒空间引入到覆盖粗糙集中,借鉴多粒度思想将其拓展为多粒度空间,通过引入对象最小描述的概念,得到一种可变精度相容粒空间的覆盖粗糙集模型。研究该模型的性质,并分析了模型中参数的不同对粒度空间精度的影响,并以此模型提出了一种约简算法。结果表明,该模型相容空间的粒度是可以通过参数调节的,粒度的可调节性使得粒空间具有伸缩性,以此模型为基础的约简算法相对于传统的约简算法提高了运算效率。

In the past, on the basis of researching and analyzing rough set approach covering,in view of a problem that the current study is limited to the field of improving the conditions of the relationship between objects, object-relational optimization to find the class, in terms of covering the space domain restrictions that are too broad, in this paper, combined the whole idea of Bayesian probability, given full consideration to the internal necessary link of conditions and results,propose a compatible tablets space quantization based on toler- ance relation. The compatibility of space is introduced into covering rough set,learn mote granularity idea to expand the space for multigranularity,by introducing the concept of minimum description of the object,obtain a covering rough set model of variable precision tolerance grain space. Research on the nature of the model,and analyze the effect of different particle size spatial accuracy of the model parameters, and thus present a reduction algorithm. The results prove that the model size is compatible with the parameter space earl be adjusted, the particle size can be adjusted so that the grains of space has scalability, the reduction algorithm based on this model,compared with the traditional reduction algorithm,improves the operational efficiency.