Research And Application Of Fuzzy Clustering Based On Tissue Like P-system

The P system is a computing model abstracted based on the communication and reaction among biological cells.The parallel computing features make it powerful computing functions.As an important data analysis technique,cluster analysis has been widely studied.It can accurately and quickly obtain valuable information from large quantities of text data,but there are still some perfections in certain details,such as initial clustering.The selection of cluster centers is not sensitive,the initial distribution of data has a greater impact on the clustering results,the selection of parameters is uncertain,the stability of the clustering results is poor,and the algorithm time complexity is high.Combining the system with the clustering algorithm can take advantage of the great parallelism function of the P system itself and reduce the time complexity of the algorithm,but the system still has a low universality and other work that needs to be further improved.Based on this,the dissertation focuses on the combination of P system and clustering algorithm,and carries out the following research work: The tissue-like P system is improved,and the cell function model capable of endocytosis and exocytosis is introduced into the tissue-like system.The tissue-like P system with endocytosis and exocytosis has been theoretically verified.The verification results show that the tissue-like P system with endocytosis and exocytosis has complete computing power.The multi-membrane structure of the tissue system and the membrane release rules and membrane engulfing characteristics of the active membrane are used to realize the generation and screening of clustering center points.Experiments have proved that the iterative process of the clustering algorithm can be effectively implemented by the system,which improves the similar organization.The universality of the P system.The paper proposes two improved clustering algorithms,one is the improved K-medoids clustering algorithm(CFKM-RADTP)of the tissue-like P system based on endocytosis and exocytosis.The other is the improved fuzzy C-means algorithm(s FCM-RADTP)based on the tissue-like system of endocytosis and exocytosis.The former optimizes the initial clustering center in one step,and changes the algorithm rules in the following.It uses the steps of preprocessing the initial clustering center proposed in Chapter 1,and realizes the use of rules and multi-stage method to adjust the algorithm.The problem of multiple scans and generation of redundant candidate center points is optimized and improved to improve the performance of the algorithm.The latter uses preprocessing rules to evaluate the initial random points,adding a step of fuzzy factor.Choose a better initial cluster center,and at the same time perform staged optimization in the selection of subsequent cluster center points.The improved algorithm is combined with the P system to improve the performance of the algorithm.Both algorithms have been tested experimentally on UCI datasets and selected artificial datasets.After comparison,the two proposed fuzzy clustering algorithms can get better clustering results in a short time.Two improved clustering algorithms will be proposed for the study of text clustering problems.The improved K-medoids clustering algorithm(CFKM-RADTP)is applied to the short text clustering problem;the improved fuzzy C-means clustering algorithm(s FCM-RADTP)is applied to the normal text clustering problem.After comparative experiments,the results are obtained by comparing the clustering accuracy and the running time of the clustering algorithm.The experimental results show that the two algorithms proposed in this paper can obtain better clustering results.