当前位置: X-MOL 学术Fuzzy Set. Syst. › 论文详情
Our official English website, www.x-mol.net, welcomes your feedback! (Note: you will need to create a separate account there.)
A fitting model for attribute reduction with fuzzy β covering
Fuzzy Sets and Systems ( IF 3.9 ) Pub Date : 2020-07-01 , DOI: 10.1016/j.fss.2020.07.010
Zhehuang Huang , Jinjin Li

Abstract As an intuitive extension of covering-based rough set models, fuzzy β covering provides an effective means to deal with uncertain information. However, classical fuzzy β covering can not fit a given data set well, which limits its popularization and application. On the one hand, this model can not guarantee that the lower approximation is contained in the upper approximation. It does not ideally describe the differences in sample classification. On the other hand, fuzzy β covering is sensitive to noise hidden in data, which leads to unstable classification performance. To address these issues, we set forth a new fitting model for attribute reduction based on fuzzy β covering. To this end, parameterized fuzzy β neighborhoods in terms of a family of fuzzy coverings are introduced to characterize the similarity between samples, using which the granular structures of fuzzy lower and upper approximations of a decision are presented. Fuzzy β covering decision tables are then formalized to deal with knowledge reduction from the viewpoint of maintaining the discriminative ability of a fuzzy covering family. Furthermore, a heuristic attribute reduction algorithm is developed to reduce redundant fuzzy coverings. Extensive numerical experiments are further conducted to examine the effectiveness and feasibility of the proposed model.

中文翻译:

模糊β覆盖的属性约简拟合模型

摘要 作为基于覆盖的粗糙集模型的直观扩展,模糊β覆盖提供了一种处理不确定信息的有效手段。然而,经典的模糊β覆盖不能很好地拟合给定的数据集,限制了其推广应用。一方面,该模型不能保证下近似包含在上近似中。它不能理想地描述样本分类的差异。另一方面,模糊β覆盖对隐藏在数据中的噪声敏感,导致分类性能不稳定。为了解决这些问题,我们提出了一种新的基于模糊β覆盖的属性约简拟合模型。为此,引入基于模糊覆盖族的参数化模糊β邻域来表征样本之间的相似性,使用它来呈现决策的模糊上下近似的粒度结构。然后从保持模糊覆盖族的判别能力的角度将模糊 β 覆盖决策表形式化以处理知识约简。此外,开发了一种启发式属性减少算法来减少冗余模糊覆盖。进一步进行了广泛的数值实验,以检验所提出模型的有效性和可行性。开发了一种启发式属性减少算法来减少冗余模糊覆盖。进一步进行了广泛的数值实验,以检验所提出模型的有效性和可行性。开发了一种启发式属性减少算法来减少冗余模糊覆盖。进一步进行了广泛的数值实验,以检验所提出模型的有效性和可行性。
更新日期:2020-07-01
down
wechat
bug