欢迎您！东篱公司
退出
-
申报数据库
-
立项数据库
-
成果数据库
-
项目获奖数据库
中文摘要：
属性约筒是粗糙集理论的重要研究内容之一，其中基于区分矩阵的约筒算法是一种高效的约简算法，但算法具有很高的空间复杂度。为了减少区分矩阵的空间开销，利用浓缩树结构，结合区分矩阵单个属性一定为核属性的特征，提出改进的生成浓缩树算法，压缩存储区分矩阵中的非空数据项，且不丢失原区分矩阵的所有信息；利用生成的浓缩树结构结合启发式策略，给出属性约简算法。实验结果表明，算法正确有效并且空间复杂度有明显降低。
英文摘要：
Attributes reduction is one of the important parts studied in rough set theory. The reduction algorithms based on discernable matrix are efficient attributes reduction algorithms, but have very high space complexity. In order to reduce the storage cost of the discernable matrix, in this paper it employed the condensing tree and combined together the speciality of the discernable matrix which single attribute must be the core attribute, proposed an improved generating condensing tree algorithm to compress and store the nonempty data item in a discernable matrix which ensures no any information will be lost in original discernable matrix; then the attribute reduction algorithm based on gener- ated condensing tree structure was provided with heuristic strategy. It was proved by the experiments that the algorithm is correct and effective. Moreover the algorithm＇ s space complexity is obviously decreased.
同期刊论文项目
同项目期刊论文
期刊信息
位置：