Two Classes Of Biological Computing And Applications In Data Mining

Membrane computing is a new branch of biological computing. It aims to getnew idea from living cells and tissues, organs, or other structures in the way of co-operation between cells. At present, there are three main types of P system: cell-likeP system, tissue-like P system and spiking neural-like P system. They are abstractedfrom cells, tissues and the nervous system, respectively. The main research directionsare: Computing power and computational efciency of diferent P system, new modelof P system, applications and implements of P system. Beneft from the parallelism,membrane computing has been applied to economics, linguistics, biological modeling,cryptography, computer graphics, and other felds.DNA computing is based on biological DNA. DNA computing belongs to the bio-chemical reactions in essence. DNA computing loads information in DNA chain. Incertain circumstances, DNA computing solves problems through DNA molecule denat-uration and renaturation annealing, copy, paste, etc. in a test tube or on the surface oron a chip. DNA computing has three distinct advantage:(1) high parallelism and fastcomputing speed (2) large capacity of storage (3) low energy consumption. At present,the main research directions are: new model of DNA computing, solving NP problemsby DNA computing, DNA automaton.Data mining is a process to obtain knowledge and information from huge data.Under the background of big data, how to analyze data efectively and discover available knowledge from vast information, improve efectiveness and availability of data, is animportant research subject of data mining. Clustering analysis plays an important rolein data mining, which is an efective way to fnd useful information. So far, manyclustering algorithms are put forward, including Hierarchical clustering, density basedclustering, grid based clustering, etc. Each have advantages and disadvantages of thesemethods. In certain areas, these methods achieve ideal efect. The paper is studied fromtwo aspects. The one is membrane computing based clustering analysis. The other isDNA computing based clustering analysis. The main work is as follows:(1)Pǎun pointed out that the exploration of new P system structure is an importantdirection in the development of membrane computing. In this paper, we extend thetraditional P system. P System with foor membrane structure, P systems with activepromoters and inhibitors, SN P system with priority and multiple output neurons areproposed respectively. Discrete Morse theory provides efective tool to analyze thetopological structure of discrete objects, and is successfully applied to image, graphictopology characteristics in the study. In view of the widespread application of discreteMorse theory, P system is combined with the structure. Latticed based P system andsimplex based P system are provided.Membrane computing is a famous approach for parallelism. It can reduce com-plexity of clustering to some degree. However, only a few studies combine membranecomputing with clustering. P systems with active promoters and inhibitors is combinedwith chameleon algorithm. The new algorithm obtains the k-nearest graphs, completethe partition of graph, aggregate subgraphs through communication rules and rewritingrules with the help of active promoters and inhibitors. The whole process of is shownby a10points test data set, which indicates the feasibility of the algorithm. We provide spiking neural P systems with priority and multiple output neurons into the applicationof spatial cluster. We use rhombic grids graph to divide data frstly, design7neuronsto stand for a rhombic grid, compare data with the threshold automatically, all theprocess of clustering implement in the new SN P system. We describe membrane struc-tures on lattice with communication rules. Clustering is implemented by supremumand infmum rules. The result is obtained through output membrane. All the processesare conducted in membranes. The new P system provides an alternative for traditionalmembrane computing.(2) Adlman-Lipton DNA computing model based topology clustering algorithm,modifed DNA sticker model based clustering algorithm, three dimensional DNA com-puting model based clustering algorithm are proposed. Adlman-Lipton DNA computingmodel based topology clustering algorithm uses single-strand DNA to stand vertex andedge, and use gel electrophoresis and sequencing technology get clustering result. Themodifed DNA sticker model is combined with clustering problems using parallel bio-chemical reactions to fnd the k-nearest graph for each cluster, cut edges beyond thethreshold, deal with outlines and noise, and obtain the cluster result. we utilize DNAcomputing using three-dimensional DNA structure(also called k-armed DNA structures)and grid tree to execute the clustering algorithm.In this paper, we extend traditional membrane computing model. P system withnew structure is also proposed. The computation completeness is proved by formallanguage. The extended model and new structure based model are used in clusteranalysis. Adlman-Lipton DNA computing model, modifed DNA sticker model andthree dimensional DNA computing model are combined with clustering as well. The newalgorithms are applied in real data set. Beneft from the high parallelism of membrane computing and DNA computing, they have great potential in data mining.