Extended P System And Its Research And Application In Cluster Analysis

Membrane computing is a new type of computing method which simulates the structure and function of cells.The model of membrane computing is often called membrane system or P system.Membrane computing,as one of the youngest branches in the natural computation,develops rapidly in the field of computing due to the biological theories and technology integration.P systems mainly consist of membrane structure,objects and evolution rules.Due to the independency of organelles and cell membrane in living organisms,P systems run in the maximum parallel mode,which have the computational power equivalent to the Turing machine and even have the possibility beyond it.Due to the combination of the Internet and traditional industries,the data of various forms has been pouring in,and the huge data processing problems that come with it have become a huge challenge in the context of big data.The features of big data are large quantities,complex structures and many types,therefore,we need to combine new calculation method,incorporate new calculation model,and improve the validity and the readability of the data.While the study of new methods and models becomes important research topic in data mining field.Cluster analysis,as an effective means to process data and extract knowledge from data,which is an important part of data mining,is widely recognized in the fields of pattern recognition and image processing.The research contents of this paper are as follows:1.The design of adaptive extended P systems based on structure and operation expansion and their calculational ability and calculational efficiency analysisFour kinds of adaptive extended P systems are put forward which extend the traditional P systems from the angle of the structure and operation:SN P systems with membrane division and dissolution,chained P systems,tissue-like P systems with cooperating rules and time free cell-like P systems with multi-promoters/inhibitors.In SN P systems with membrane division and dissolution,the function of redundant cell removal is realized by introducing membrane dissolution rules which can reduce the use of resources.In chained P systems,the system structure,objects,and rules are all chained.In tissue-like P systems with cooperating rules,rules of each membrane are divided into different components,and each component is assigned a number.Only one component is active when rules execute,that is,only rules in the active component have the chance to execute.In this way,rules within the same component can work together to complete a function.In time free cell-like P systems with multi-promoters/inhibitors,each rule can have multi-promoters/inhibitors,and arbitrary promoter/inhibitor can achieve the function of promotion/inhibition.At the same time,the calculational abilities of the extended P systems are proved by the simulation of the register machine,matrix grammar and Lindenmayer system,and the calculational efficiency of the extended P systems are analyzed by solving the NP problems.2.The general object-oriented description,modeling and reusable components design for P systemsThe concepts in P systems are summarized into six classes:P system object,membrane,rule,cell-like P system,tissue-like P system and neural-like P system,and the object-oriented description are given.A data structure of P systems is designed.The object-oriented static model and dynamic model are designed which summarize the static description and the dynamic implementation of P systems.At the same time,the reuse components are built,the general components and the areas components are extracted.The collection of components with maximum particle size is built.These components are embedded into the P systems implementation platform in the manner of plug-ins and buttons.3.Design and application of a variety of improved cluster algorithms based on the extended P systemsImproved cluster algorithms based on the direct membrane algorithms are con-structed which combine the extended P systems with cluster algorithms,including:DB-SCAN algorithm based on time free cell-like P systems with multi-promoters/inhibitors,ROCK algorithm based on chained P systems,the graph cluster algorithm based on the chained P systems and the consistent cluster algorithm based on time free cell-like P systems with multi-promoters/inhibitors.The algorithm rules are designed,the algo-rithm flows and the algorithm complexities are analyzed,and the algorithms are used to data set to verify the availability and effectiveness.The experiments show that the efficiency and the precision of the algorithms based on the direct membrane algorithms are improved.At the same time,a PAM algorithm based on time free cell-like P systems with multi-promoters/inhibitors is designed,which is used to solve the cluster problem on a data set consists of 45 real news headlines.The availability and effectiveness of the proposed algorithm is verified.As a new calculational method,the proposal of extended P systems and their systems analysis are important research contents in membrane computing.At the same time,membrane computing has been preliminarily applied in many fields,however,the applications of cluster analysis are fewer.Therefore,the extended P systems and their applications in cluster analysis are of great scientific significance and practical value.