Research And Application Of Clustering Algorithm Based On Dynamic Coupled Tissue-like P Systems

In recent years,the development speed of information technology and data update speed are accelerating.The data sources are complex and diverse,and the amount of data has increased rapidly at an unprecedented rate.In order to obtain the valuable information quickly and accurately from the huge data,cluster analysis is an important data analysis technology that has been widely studied.But there are still some disadvantages such as initial cluster center and parameter selection sensitivity,data distribution on a greater influence on the clustering results,poor stability of clustering results,and high algorithm time complexity and so on.The P system is a computational model abstracted from the interactions and reactions between living cells.It has powerful computational power as the same as Turing machine.By combining the P system with the clustering algorithm,utilizing the maximum parallelism of the P system reduces the time complexity of the algorithm,but the P system has the disadvantages of poor universality and so on.This thesis studies the tissue-like P system,the fuzzy c-means clustering algorithm and the spectral clustering algorithm.The main research contents are as follows:1.In this thesis,we introduce the research background and significance,and the domestic and foreign research present situation of tissue-like P systems,the fuzzy c-means clustering algorithm,spectral clustering algorithm and clustering algorithm based on tissue-like P systems.Meanwhile,the basic theory of the tissue-like P system,fuzzy c-means clustering algorithm,and spectral clustering algorithm are expounded.2.The tissue-like P system was improved by introducing coupled cells and cells that they can undergo division and dissolution.The dynamically coupled tissue-like P systems were proposed.This system can realize the iterative process of clustering algorithm and improve the universality of P systems.3.The improved fuzzy c-means clustering algorithm based on the dynamic coupled tissue-like P systems and the improved spectral clustering algorithm based on the dynamic coupled tissue-like P systems are proposed.The former introduces the density idea into the fuzzy c-means clustering algorithm to find the initial clustering center,and uses the improved gaussian kernel function as the distance metric formula to optimize the fuzzy c-means clustering algorithm.We combine the optimized algorithm with the P systems to improve the computational efficiency of the algorithm.The latter introduces the neighborhood idea into the spectral clustering algorithm to improve the similarity measurement method,and then uses the improved fuzzy c-means clustering algorithm proposed in this thesis to cluster the obtained feature vectors.Same as above,we combine the improved algorithm with the P systems.These improved clustering algorithms were tested on the artificial datasets and the UCI datasets respectively,and the experimental results show that the improved two algorithms can achieve good clustering results.4.The two algorithms proposed in this thesis are used to solve the problem of image segmentation.The improved fuzzy c-means clustering algorithm is used for image segmentation of color images,and the improved spectral clustering algorithm is used for image segmentation of real images.The segmentation results show that the two clustering algorithms can obtain better segmentation results.