测试代价敏感的粗糙集方法

发布时间：2018-09-11 08:50

【摘要】：在粗糙集模型中,α量化不可分辨关系是强与弱不可分辨关系的推广形式.然而值得注意的是,基于这三种不可分辨关系的粗糙集并未考虑数据中属性的测试代价.为解决这一问题,提出了测试代价敏感的α量化粗糙集模型,从二元关系的角度使得粗糙集模型代价敏感,并将新模型与基于强不可分辨、弱不可分辨以及传统α量化不可分辨关系的粗糙集模型进行了对比分析.进一步地,通过分析传统启发式算法在求解约简的过程中未考虑降低代价这一不足之处,提出一种新的属性适应性函数,并将其应用于基于遗传算法的约简求解中.实验结果表明该方法不仅可以降低由边界域所带来的不确定性而且同时降低了约简后的测试代价.
[Abstract]:In rough set model, 伪 -quantization indiscernibility relation is a generalized form of strong and weak indiscernibility relation. It is worth noting, however, that the rough set based on these three indiscernible relationships does not take into account the test cost of attributes in the data. In order to solve this problem, a test cost sensitive 伪 quantization rough set model is proposed, which makes the rough set model cost sensitive from the point of view of binary relations, and the new model is based on strong indiscernibility. The rough set model with weak indiscernibility and traditional 伪 -quantization indiscernibility relation is compared and analyzed. Furthermore, a new attribute adaptive function is proposed by analyzing the shortcomings of the traditional heuristic algorithm in the process of solving the reduction without considering the cost reduction, and it is applied to the reduction solution based on genetic algorithm. The experimental results show that the proposed method can not only reduce the uncertainty brought by the boundary domain, but also reduce the test cost after reduction.
【作者单位】：南京大学工程管理学院;江苏科技大学计算机科学与工程学院;南京大学智能装备新技术研究中心;南京大学软件新技术国家重点实验室;
【基金】：国家自然科学基金(61572242,71671086,61473157) 江苏省普通高校研究生科研创新计划项目(KYLX16_0021)~~
【分类号】：TP18

【相似文献】