流计算模式下概率粗糙集三支决策的快速计算

概率粗糙集三支决策是不确定问题求解的一种重要理论,流计算模式是一种新型的动态内存计算形式,实施流计算模式下三支决策的快速动态计算是一项具有挑战性的新议题。本研究以流计算模式中的两个核心计算步骤即动态增量与动态减量作为研究对象,提出了一种流计算模式下概率粗糙集三支决策域的快速动态学习方法。首先对流计算模式中三支决策动态增量和动态减量的不同变化情况进行了数据建模。然后基于不同数据变化情况分别讨论了数据增量与数据减量时三支决策域的变化推理,并且基于上述理论给出了流计算模式下的三支决策动态增减学习算法。该算法能够以更低的时间复杂度获得与经典三支决策算法相同决策效果。最后通过八种UCI数据集的实验证明了流计算模式下三支决策动态增减学习算法在时间消耗上明显优于经典概率粗糙集三支决策算法,并且在不同阈值下具有稳定的决策效率。本研究表明了流计算模式下三支决策快速计算是可行的。

Fast computing of probabilistic rough set three-way decision in stream computing mode

Probabilistic rough set three-way decision is an important theory for solving uncertainty problem, stream computing mode is a new form of dynamic memory data computing, it is a challenging topic to carry out dynamic study research of probabilistic rough set three-way decision in stream computing mode. Taking the two core computing steps of stream computing mode, that is, dynamic increment and decrement, as the research object, a fast dynamic learning method is proposed for three-way decision region of probabilistic rough set in stream computing mode. Firstly, the data modeling of different changes of dynamic increment and dynamic decrement of three decision-way decision in stream computing mode was carried out. Then the reasoning of change of three-way decision region in data increment and data decrement is discussed respectively based on different data changes. Based on the above theory, a three-way decision dynamic incremental and decremental learning algorithm is given in stream computing mode. The algorithm can obtain the same decision effect as the classical three-way decision algorithm with lower time complexity. Finally, the experiments of eight UCI data sets proved that time efficiency of three-way decision dynamic incremental and decremental learning algorithm are superior to the classical probabilistic rough set three-way decision algorithm in stream computing mode, and has stable decision efficiency under different thresholds. This research proves that the fast computing research for three-way decision is feasible in stream computing mode.