Research Of Cell-like Membrane System In Clustering Algorithm

Membrane computing is a new method of calculation. The researchers were inspired fromthe structure and function of living cells, thereby abstracting this computing model. Itscharacteristics of distribution, parallelism and non-determinism, make it have more advantagesin the research of computing science. Now it should be no wonder that most classes of P systemsconsidered so far are computationally universal, equal in power with Turing machines and it ispossible to show the capacity of exceeding the power of Turing machines. So it has greatsignificance in solving concurrency problems and optimization problems and now a variety ofapplications have been reported, such as biology, medical science, computer graphics, linguistics,economics, sociology and computer science. Currently, experts and scholars in various fieldshave attracted more and more attention on the membrane computing, which make it a hotspot inAcademia at home and abroad.With the development of current society and information industry, the research of datamining has attracted considerable attention of all society. Clustering analysis, as an importantbranch of data mining, is an important means of solving the problems of data mining. Clusteringanalysis is the process of discovering specific data distribution and dividing the sets of samplesinto similar categories. It has been widely applied in many areas, including neural networks,image processing, modern biology and statistics.Considering the non-determinism and parallelism of membrane computing, we introduce itinto clustering algorithms and make it possible to expand the scale of data sets and increase thecomputing speed. Three main contents are contained in this work. Firstly, based on the featuresof K-AGNES algorithm and the advantages of membrane computing, a P system is proposed torealize K-AGNES algorithm. By inputting dataset of n objects, object matrix and cluster number,we design the membrane structure, objects, evolution rules and priority relation, and finallyobtain k clusters of n objects after operation. In order to solve the complex problem ofspatial query in traditional density-based clustering algorithm, the P system is first applied to thedensity-based clustering method to realize DBSCAN clustering algorithm, which is aninnovation for application of membrane computing. The P system with active membranes is aspecial P system. Its division rule has a great advantage of generating an exponential workingspace in a linear time. This paper proposed a new P system with active membranes to solve DBSCAN clustering problems. This new model of P system can reduce the time complexity ofcomputing without increasing the complexity of the DBSCAN clustering algorithm.With the Popularity of the Internet, e-commerce systems have more influences commodityeconomy and people’s daily economic life. Along with the rapid development of e-commerce andthe growth of online products, its structure has become more complex; customers could notprecisely find their own needs in wide range of goods successfully. For this reason, variousrecommendation systems came into being. On the basis of the current situation of e-commercewebsites and the existing problems in, a commodity recommendation system can be regarded asa weighted and undirected graph and can be transformed into a density-based clustering problem.Then this paper proposes a P system with active membranes based on DBSCAN clusteringalgorithm to solve commodity recommendation problems. This new method of commodityrecommendation can partly improve the purchasing rate and enhance the competitiveness of theenterprises. The success of solving this problem by the P system is helpful to conduct furtherresearch of membrane computing in practical issues.