Research Of Cell-like P System In Partitional Clustering

Membrane computing is a new natural method of calculating. It aims to achievecalculation process by simulating the function of living cells, tissues and organs.Objects in this model, which has complete computing capability, can evolve in amaximal parallelism and distributed manner. In recent years, membrane computingmodel has been applied to practical problems of biology, medicine, computer science,economics and so on. It is a promising area of research, and is expected to play a keyrole in the new field of computing technology. Theoretically, some simple membranesystem already has the same computing capability of Turing machines, given itspowerful computing capabilities, membrane computing can exceed Turing machine,or even replaced. Therefore, the study of membrane computing has received scholars’well attention.As a very useful tool in data mining techniques, cluster analysis has drawnincreasing attention. Cluster analysis is a common tool for data analysis, and it hasbeen widely used in many application areas, including market research, patternrecognition, data analysis, image processing, and so on. With the data growing at anunprecedented rate, society has entered the era of big data. Data is the wealth ofknowledge, and how to use these large-scale data is particularly important. Clusteranalysis is an important human activity, it is a process that divides data into clusters(intra-cluster data are similar and inter-cluster data are dissimilar). It can help peopledig out new business opportunities from these massive data. Thus, cluster analysiswill become a very active research topic in data mining.To handle these massive amounts of data, the existing computers are increasinglydifficult to meet human needs, so scientists began looking for a new computing model,hoping to improve the speed of the existing computer. Based on P systems’ greatparallelism, the distributed manner and so on, this paper attempts to combine Psystems with clustering algorithm. Under the premise of clustering quality, using membrane computing to improve the speed of clustering operation to help peoplebetter cope with big data era.The main work of this paper: Firstly, The K-medoids clustering Algorithmbased on cell-like P system was proposed, which was a new attempt of application inmembrane computing, It uses distance to define dissimilarity of any two objects anddifferent membranes to represent different clusters, and objects enter into themembrane of the smallest dissimilarity, until no change in the center of eachmembrane; Secondly, a class of cell-like P system based on the MapReduce structureis proposed in this paper for their commonalities, in this P system, a series ofmembranes represent workers of the map and reduce stage of the MapReduce, objectsin various membranes evolve according to certain rules, and the final objects in theoutput membrane shall be the processing results of MapReduce; Thirdly, a class ofcell-like P system based on the MapReduce structure and K-means clusteringalgorithm are combined, which provides another new algorithm, according to thecharacteristics of the P system, as well as the core idea of K-means clusteringalgorithm, the proposed new algorithm can complete the clustering process in ashorter time under the premise of clustering quality assurance.