| Data Mining is a ordinary proceduce that taking effective and potential value of new information and knowledge from quantities of data.In many of the mining methods,Rough Set methods is one of the useful tools of Data Mining.Rough set theory is a mathematical method to deal with vague and uncertain question which proposed by Pawlak in the early 1980s and can make a analysis of the rule contained in the data without any additional information.Rough set theory gets success in many fields,such as pattern identifying, knowledge discovery, machine learning and decision analysis.The attribute reduction of information systems is one of the important content in rough set theory. To find the best reduction or all reduction of information systems is the NP-hard problem.Genetic algorithm is a optimization method which simulates the process of biological evolution,with global search, robustness,scalability, implicit parallelism, etc.,and has been widely used to the attribute reduction of Rough set.To get the minimal reduction of the information systems quickly and effectively, Based on genetic algorithm and analysis of attribute reduction,this paper looks on the significance of attributes as heuristic information introduces the heuristic information into genetic algorithm, and proposes an improved heurisitic genetic algorithm.This paper proposes a new mutate operator for introducing the heurisitic information and the technology is a local search method using heurisitic information.Therefore,the algorithm has global optimizing ability and converges quickly. Differential evolution algorithm is a global evolutionary optimized algorithm,which based on mutation,crossover,selection. This algorithm can applied to a number of individuals so that most of them can optimized from generation to generation, and finally it gradually approach the optimal solution.This paper will introduce differential evolution algorithm in attribute reduction of rough set,and construct a differential evolution algorithm which based on attribute reduction of rough set.The results show that the two methods can calculate minimal reduction of decision charts quickly and effectively. |
