决策表属性约简集的增量式更新算法

针对动态变化的决策表,研究了属性约简集的动态更新问题。在详细分析新增对象的所有可能情况的基础上,提出一种基于分辨矩阵元素集的属性约简集增量式更新算法。该算法根据新增对象的不同情况快速更新分辨矩阵元素集,依据分辨矩阵元素集中增加和减少的元素有效地更新原属性约简集,快速得到新的最小属性约简。最后,通过5个UCI的数据集验证了算法的有效性。

An Incremental Updating Algorithm for Attribute Reduction Set of Decision Tables

The dynamic updating algorithm of attribute reduction set is discussed in the dynamic decision tables. Some different situations are analyzed when new objects are added to the old decision table, and an incremental updating algorithm for attribute reduction set based on the discernibility matrix element set is proposed. In the algorithm, the discernibility matrix element set could be updated quickly. Thus, the old attribute reduction set can be updated effectively by the changes of discernibility matrix element set, and the new minimum attribute reduction can be got quickly. At last, the correctness and effectiveness of the algorithm on 5 real datasets are demonstrated.