摘要
属性约简是粗糙集理论的核心问题之一,寻找最优约简现已证明是NP-hard问题。本文在差别矩阵的基础上,采用贪心算法的策略,提出了一种启发式算法来寻求最优约简。该算法不仅能在有核的决策表中通过核进行属性扩充求得最优约简,而且在无核的决策表中也能根据属性在差别矩阵中出现的频率求得最优约简。经实例验证该算法是有效的。
Attribution reduction is one of the key problems on rough＇set theory. Nowadays searching the optimal attribute reduction of the information system has been proposed the NP - hard problem. Based on the conjunction discernible matrix takes the strategy of the greedy algorithm we propose a kind of heuristic algorithm to get the optimal reduction in this paper. This algorithm can obtain the optimal reduction through the expansion of the core attributes in which there are cores of the decision table. Moreover in non - core decision table, it can also act according to this attribute the frequency which appears in the conjunction difference matrix to obtain optimal reduction. All experiment shows that the algorithm is highly effective.
出处
《忻州师范学院学报》
2008年第5期11-13,共3页
Journal of Xinzhou Teachers University
基金
忻州师范学院院级科研基金项目(项目编号:N200504)
关键词
决策表
属性约简
差别矩阵
decision table
attribute reduction
conjunction discernible matrix
