A Class Of Clustering Algorithms Combined With Membrane Computing And Immune Mechanisms

Data clustering technique has been applied in many areas extensively. The cluster analysis to the data objects is one of the main research respects of the cluster analysis techniques.The artificial immune systems developed on the basis of researching and making use of the various principles and mechanisms of biological immune systems. The artificial immune systems are a variety of multi-class of computing systems technology, information processing technology and various scientific applications collectively. Membrane computing, known as P system, is a novel kind of computing models. Membrane computing has several attractive features which have been applied on the clustering algorithm, such as distribution, maximum parallelism and so on.This thesis proposes two clustering algorithms which combine membrane computing with immune mechanisms to solve the clustering analysis problems. The algorithms are in the framework of membrane computing, inspired by the immune system and clone selection mechanism of the artificial immune system and combine with the existing research results. The proposed clustering algorithms are estimated on four data sets respectively. The experimental results compared with other clustering algorithms conduct a comparative analysis. The main research contents are as follows:(1) A clustering algorithm based on immune mechanisms and membrane computing(shorted for IM-MC). To evolve the objects, the algorithm puts the three operators(selection, crossover, mutation) of the immune system into the clustering algorithm which is in the frame of the membrane computing. It uses the transportation mechanism of the P system to enjoy the excellent objects among membranes to search the cluster centers globally. The algorithm is estimated on three artificial data sets and one real-life data set. The experimental results are compared with k-means, GA-based, PSO-based and conduct a comparative analysis result.(2) A clustering algorithm based on clone selection mechanisms and membrane computing(shorted for CSA-MC). To evolve the objects, the algorithm puts the three operators(clone, mutation, selection) of the clone selection mechanisms into the clustering algorithm which is in the frame of the membrane computing as the evolution rules. The algorithm uses transportation mechanism of the P system to share the excellent objects. This algorithm is estimated on the above four data sets respectively. The experimental results are compared with other two algorithms and conduct a comparative analysis.