Abstract: Because it is easy for clustering algorithm based on Particle Swarm Optimization to fall into the local optimum，clustering of gene expression data using Quantum-behaved Particle Swarm Optimization is proposed.The control of diversity of particles is applied in the novel clustering algorithm to improve the global convergence.A K-means operator like in K-means clustering is also introduced to accelerate the convergence of proposed algorithm.Rand index and Silhouette index are selected as evaluation criteria of clustering.Clustering experiment on five artificial or real gene expression data sets shows that the new method outperforms the PSO clustering on convergence speed and the global convergence is proved through the control of diversity.Contrast to some common clustering algorithms，the better clustering solution and validation are also obtained.

Key words: Quantum-behaved Particle Swarm Optimization, gene expression data, diversity-guided, clustering

摘要： 针对基于粒子群优化的聚类算法容易陷入局部最优值的缺点，提出将量子行为粒子群优化应用于基因表达数据的聚类分析问题中。在新的聚类算法中采用了对粒子群的多样性控制，以提高算法的全局收敛性能；此外还在新算法中引入了类似于K均值聚类的操作步骤，用以提高算法整体的收敛速度。选择Rand指数和Silhouette指数作为聚类评价标准，对5个人工和实际的基因表达数据集合进行聚类实验分析表明，新算法和基于粒子群优化的聚类算法相比，具有较快的收敛速度，粒子多样性的控制能有效改善算法的全局收敛性能。和其他一些常用的聚类算法比较，也能够获得更好的聚类评价，聚类效果更好。

关键词: 量子行为粒子群优化, 基因表达数据, 多样性引导, 聚类