论文部分内容阅读
通过分析现有分辨矩阵的不足,给出了基于条件熵的分辨矩阵定义,以得到不一致决策信息系统下的合理条件属性约简结果.基于改进分辨矩阵的定义,提出了相应的条件属性约简算法,并证明了其约简为不同于代数观点下的约简,而与信息观点下的约简相同.进一步,对该改进分辨矩阵做进一步修正,提出了一个增量式条件属性约简算法.实验结果表明:所提出的改进分辨矩阵下的条件约简算法总体上优于同类求解算法,而其增量条件约简算法能够获得稳定可靠的约简结果,同时约简效率得到较大的提高.
By analyzing the deficiency of existing discernibility matrices, the discernibility matrix definition based on conditional entropy is given to obtain the reasonable condition attribute reduction results under the inconsistent decision information system. Based on the definition of improved discernibility matrix, the corresponding condition attribute reduction Algorithm and proved that its reduction is different from the reduction under the algebraic viewpoint and the same as the reduction under the information viewpoint.Furthermore, the improved discernibility matrix is further amended and an incremental conditional attribute reduction algorithm The experimental results show that the proposed condition reduction algorithm based on improved discernibility matrix is better than that of the same solution algorithm in general and the incremental condition reduction algorithm can obtain a stable and reliable reduction result while reducing the efficiency of the reduction improve.