An Attribute Reduction Of Rough Set Based On PSO

Rough set theory is proposed to deal with uncertainty and incomplete information provides powerful mathematical tools. Minimal attribute reduction is an important content of rough set theory, it is to eliminate redundant information, the necessary steps in rules. But minimal attribute reduction is a NP-hard problem solving, traditional reduction algorithm is suitable for processing the data of low dimension small minimum attribute reduction problem, as the number of attributes, it increases the computational complexity is exponential growth. Therefore, many of the improved algorithms are put forward. In part of the improved algorithm is the greedy algorithm, when the solution space rather complicated, it is easy to fall into local optimal solution. The other part is the heuristic algorithm, even the genetic algorithm which is good for global search, the benchmark of global search and easy to appear the precocious phenomena, and urge us to find a kind of fast and effective attribute reduction algorithm to solve the problem.PSO optimization algorithm is a kind of heuristic search algorithm, and it has the same global search function, as genetic algorithm. Because the algorithm is simple, easy to realize, it has been successfully applied to the artificial neural network training methods, the function optimization and optimization, the mini/max problem, the multi-objective optimization problems are obtained in the application of success.PSO algorithm mainly adapted continuous space function optimization, if the space of little change, also can be PSO algorithm is applied to discrete optimization problems. this paper studied and designed a kind of PSO is based on rough set attribute reduction algorithm : Through the attribute binary coding, attribute dependence to define Fitness function, PSO optimization arithmetic, and binary decoding corresponding properties to achieve the attribute reduction. This algorithm is introducing the computer cache to reduce the complexity of the algorithm, which is simple to use, and have good advantage. MATLAB simulation results show that this method can effectively obtain rapid, attribute reduction, when the number of attributes, more than its efficiency.So, for the solving of Minimal attribute reduction under the complex environment, compared with other algorithms, the standard PSO attribute reduction algorithm is effective.