基于粗糙集的属性约简和核的快速更新算法分析-analysis of fast updating algorithm of attribute reduction and kernel based on rough set.docxVIP

优秀毕业论文 精品参考文献资料 In this dissertation, the research trends and theoretical knowledge of rough set are Presented firstly. Then the existing attribute reduction methods, which consists of Pawlak algorithms, algorithms based on discernibility matrix, algorithms based on information entropy. The dissertation is mainly studying and learning from existing research, the new ideas and contributions included in this dissertation are summarized as follows: A quick improved algorithm of attribute reduction (Pawlak Reduction) based on positive region is proposed which make use of bitmap and granular computing. In this algorithm, a new heuristic information which can efficiently reduce the numbers of granular computing is proposed. That is to say, some computations which can not change the results of attribute reduction are reduced. The emulate example and the experimental results show that the proposed algorithm is more efficient and more reliable. The definition of attribution reduction of simplified binary discernibility matrix is provided. At the same time, it is proved that the above definition of attribution reduction is the same as the definition of attribution reduction based on information entropy. In order to compute simplified binary discernibility matrix, a quick algorithm for simplified decision table is adpoted. On this condition, a quick attribution reduction algorithm based on simplified binary discernibility matrix of Information Entropy is designed, the time complexity and space complexity of the new algorithm are max{O(| C ||U |), O(| C |2|U′ |2 )} and max{O(| C || U′ |2 ), O(| U |)}  respectively. At last, an emulate example illustrates the new algorithm is more efficient than the typical algorithms. Some drawbacks were analyzed in some existent the computing core algorithms, in order to improve the efficiency of the dynamically updating algorithm of computing the core in decision table, At fist, the concept of simplified decision table is int