Study on dynamic attribute reduction based on improved PSO algorithm

Attribute reduction is one of the important topics in the research on rough set theory. In confrontation with dynamic data, common methods of attributes reduction have such disadvantages as unstable reduction results, intensive computation and the difficulty in meeting the need of real-time processing. To solve these problems, a method of dynamic attributes reduction with improved PSO algorithm is proposed based on the research of particle swarm optimization. The concrete work is the following: firstly, the traditional PSO algorithm is improved to enhance the global search ability, which increases the diversity of particle population distribution. Secondly, the information decision system data extraction is divided into some subdecision tables, which are reducted by used of improved PSO algorithm. Finally, each reduction results are intersected and get the most minimal reduction. Simulation and experimental results show the dynamic reduction algorithm can overcome the shortcomings of common attribute reduction which possesses the significant effect and rapid computation.