细胞型膜系统在聚类算法中的研究

发布时间：2018-08-03 08:18

【摘要】：膜计算是一种新型的计算方式，研究者从活细胞的结构和功能中获得灵感，从而抽象出这种计算模型。膜计算具有分布式、不确定性、最大并行计算等独特特征，，使其在现有的计算科学研究中显示出突出的优势，现已证明很多膜计算模型具有与图灵机等价的计算能力，甚至具备超越图灵机的可能，因此膜计算在解决一些并行问题、优化问题方面，具有很重要的研究意义，并在生物学、医学、计算机图形学、语言学、经济学、社会学、计算机科学等多个领域有着广泛的应用。目前各个领域的专家学者对膜计算的专注度越来越高，使其成为全球学术界的研究热点。随着信息产业界的发展和当前社会的不断进步，数据挖掘领域受到了全社会各个领域的极大关注。而作为数据挖掘重要分支的聚类分析，是解决数据挖掘问题的主要手段。聚类分析就是从大量的样本对象中，发现不同的对象分布情况，从而将相似的对象划分成类的过程。目前聚类分析方法已经在很多不同领域得到应用，例如神经网络、图像处理、现代生物学、统计学。基于膜计算的非确定型和最大并行性等特点，本文将其引入聚类算法中，借助P系统模型保证聚类质量的同时提高数据集规模和运算速度，主要研究内容包括以下三个方面：一是根据K凝聚层次聚类算法的特点，结合膜计算的优势，提出了一种基于细胞型P系统的K AGNES算法，通过输入具有n个对象的数据集合、对象集合矩阵及聚类个数k，结合P系统的特性，来对P系统的膜结构、膜内对象、膜内规则及规则优先关系进行设计和构建，最终得到n个对象的k个分组。二是为解决了传统的基于密度的聚类算法进行区域查询的繁琐复杂问题，构建了一个细胞型P系统来实现DBSCAN算法。将膜计算应用在基于密度的聚类算法的实现上，是膜计算应用的一个创新之举。活性膜P系统是一种特殊的P系统，细胞膜的分裂规则使其能够为计算提供指数个计算空间，具有解决聚类问题的独特优势。本文结合DBSCAN算法的特点和活性膜P系统的优势，提出了一种基于活性膜P系统DBSCAN算法，作为第三个研究重点，使其能够在更短的时间内完成聚类过程，提高聚类算法的效率。随着互联网的普及，电子商务系统对商品经济的发展和消费者的日常经济生活，产生了翻天覆地的影响。但随着电子商务的飞速发展，商品种类和数量的急剧增长，电子商务结构也变得越来越复杂，顾客很难在电子系统繁多的商品存储中，精确地找到自己想要的商品，于是电子商务推荐系统便应运而生。根据目前电子商务网站的现状及商品推荐所存在的问题，本文商品推荐系统看作一个无向加权图并将其转换成DBSCAN聚类问题，最后使用活性膜P系统来实现。这种全新的商品推荐方法在一定程度上提高顾客的购买率，增强企业的竞争力。商品推荐问题的成功解决，也将有助于膜计算在现实应用方面进行更深入的研究。
[Abstract]:Membrane calculation is a new method of computing. Researchers derive inspiration from the structure and function of living cells, thus abstracting this calculation model. Membrane calculation has the unique features of distributed, uncertain, and maximum parallel computing, which shows prominent advantages in the current research of computational science. Many membrane computing models have been proved. With the computing power equivalent to the Turing machine, and even the possibility of surpassing the Turing machine, membrane computing has a very important research significance in solving some parallel problems and optimizing problems. It has extensive applications in many fields, such as biology, medicine, computer graphics, linguistics, economics, social science, computer science and so on. Experts and scholars in various fields have paid more and more attention to membrane computing, making it a research hotspot in the global academic field.
With the development of the information industry and the continuous progress of the current society, the field of data mining has attracted great attention from all fields of society. As an important branch of data mining, clustering analysis is the main means to solve the problem of data mining. The process of dividing similar objects into classes has been applied in many different fields, such as neural networks, image processing, modern biology, statistics.
Based on the characteristics of uncertainty and maximum parallelism of membrane computing, this paper introduces it into clustering algorithm, with the aid of P system model to ensure the quality of clustering and improve the size and speed of data sets. The main research contents include the following three aspects: first, according to the characteristics of the clustering algorithm based on K, combined with the advantages of membrane computing, it is proposed. A K AGNES algorithm based on cellular P system is introduced. By input of data sets with n objects, object set matrix and cluster number k, combined with the characteristics of P system, this paper designs and constructs the membrane structure of the P system, the inside object, the rules of the membrane and the rule priority relations, and finally obtains the K grouping of the n objects. Two is to solve the problem. The traditional density based clustering algorithm is a complicated and complicated problem of regional query. A cell type P system is constructed to implement the DBSCAN algorithm. The application of membrane computing to the implementation of density based clustering algorithm is an innovative approach to the application of membrane computing. The active membrane P system is a special P system, the cell membrane splitting rule makes It can provide an exponential computing space for computing, and has a unique advantage to solve the clustering problem. In this paper, based on the characteristics of the DBSCAN algorithm and the advantages of the active membrane P system, a DBSCAN algorithm based on the active membrane P system is proposed. As the third research focus, it can complete the clustering process in a shorter time and improve the clustering algorithm. Efficiency.
With the popularity of the Internet, the electronic commerce system has a great impact on the development of commodity economy and the daily economic life of consumers. However, with the rapid development of electronic commerce and the rapid growth of commodity types and quantities, the structure of electronic commerce has become more and more complex, and it is difficult for customers to store the various kinds of electronic goods. According to the present situation of the e-commerce website and the problems existing in the recommendation of the commodity, this article is regarded as an undirected weighted graph and converted it into a DBSCAN clustering problem. Finally, it is realized by using the active membrane P system. The method of commodity recommendation improves the customer's purchasing rate to a certain extent and enhances the competitiveness of the enterprise. The successful solution of the problem of commodity recommendation will also help to make more in-depth research on the practical application of membrane computing.
【学位授予单位】：山东师范大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TP38;TP311.13

【参考文献】