Research On Quantum Genetic Clustering Algorithm Based On Variable Length Chromosome

Quantum genetic algorithm is a product of quantum computing theory combined with geneticalgorithm, which is a research field developed in the late90’s. Quantum genetic algorithm has theadvantages of fast convergence speed, strong global search ability, and it has a strong advantage indealing with optimization problems. Data mining is a new subject formed with the development ofdatabase technology. Through the data mining we can find useful data from large volumes of data.Thus, data mining has a strong practical. Cluster analysis is a kind of pattern recognitionunsupervised technique and it is a very important branch of data mining. It has been widely used invarious industries. K-means clustering is cluster analysis method, has the advantages of simplealgorithm, fast convergence speed etc. But the traditional K-means algorithm has the shortcomingsof initial value sensitivity and easy to fall into local minimum, and the clustering center number kshould be determined by experience, leading to the classification results may not be optimal.Based on the summary of previous research results, the paper presents a variable lengthchromosome genetic quantum algorithm. First of all, we make improvements to the population. Thechromosome length of the population is no longer a fixed value, but the value is in a certain range.The set of clustering center consist of variable length chromosome. This avoids the problem that theK value is determined by human experience. Secondly, we design a variable length operationfunction. In the process of evolution, the optimal clustering center is the goal to the cluster centerfor quantum rotation gate operation, at the same time to adjust the length of chromosome, in orderto realize the clustering number to change.The algorithm uses the Matlab to program, and we do experiment on different data sets, andcompared with the quantum genetic clustering algorithm and k-means algorithm. The experimentresults show that this algorithm has better clustering results. At the end of the paper, the algorithm isapplied to the customer classification. According to the results, we puts forward different marketing,and confirm the practicability of the algorithm.