Abstract: A discretization algorithm of rough set continuous attribute based on improved particle swarm optimization is proposed.For optimization of algorithm，Improved particle swarm optimizaiton is utilized.In order to overcome some defects with PSO，Initialization of the population and the inertia weight are adjusted，the diversity ability of particle swarm optimization is improved.For rough set attribute discretization，it is mainly through minimum break point sets as optimization goals and rough set attribute dependence as constraints.The simulation results show that the proposed method can effectively solve the discretization of continuous attributes in decision table and the proposed method is of high calculation speed and good convergence.

Key words: improved particle swarm optimization, rough sets, continuous attribute discretization

摘要： 提出一种基于改进粒子群优化的连续属性离散化算法。在算法优化方面，采用改进粒子群优化算法。为了克服传统粒子群优化的不足，对种群初始化和自适应调整粒子的惯性权重，提高了粒子群优化算法的全局寻优能力。在粗糙集属性离散化方面，主要是通过将最小断点集作为优化目标，粗糙集属性依赖度作为约束条件。仿真结果表明，该方法能有效地解决决策表连续属性离散化问题，计算速度快，收敛性好。

关键词: 改进粒子群优化, 粗糙集, 连续属性离散化