Research On Attribute Reduction Method Of Rough Set Based On Intelligent Optimization Algorithm

Rough set as a mathematical tools to deal with imprecise,incomplete and uncertain data,and its advantage is to remove redundant attributes can be maintained without changing the classification and decision-making, with its own advantages has been successfully applied in data mining,pattern recognition,machine learning and other fields.Attribute reduction is one of the core knowledge discovery rough set, but solving minimal reduction has been proved to be a NP-hard problem due to the combinatorial explosion problem of the attributes. Reduction algorithm introduced with random strategy become an inevitable trend to improve the current reduction algorithm.The cross, mutual penetration and mutual promotion between life science and engineering science is a significant feature of the development of modern science and technology.Intelligent optimization algorithms by simulation or reveal some natural phenomenon or process evolved, it has global, efficient parallel to optimize performance, robustness,versatility,no problem specific information and so on. In order to solve complex problems and provide a new way of thinking means,In this paper, rough set and intelligent optimization algorithms based, object attribute reduction, attribute reduction algorithm based on intelligent optimization method was studied. This paper completed the following work:1.In this paper, we mainly studies the rough set based on attribute importance degree,attribute dependency,distinguish matrix attribute reduction algorithm, and combining rough set with intelligent algorithm of attribute reduction algorithm, analysis and compare the advantages and disadvantages of existing attribute reduction algorithm.2.In view of the standard genetic algorithm for attribute reduction appear prone to premature and convergence speed at the late period of slow problem, this paper proposes an improved adaptive genetic attribute reduction algorithm(IAGA), First, the algorithm will be added to the initial population attribute core algorithm, avoiding the blindness of randomly generated initialization population. Second, adaptive crossover and mutation probability reset based, dynamic adjustment of crossover and mutation probability according to the fitness function, optimizing the ability of an individual tobe selected, increasing the diversity of the population, to avoid the premature phenomenon.3.Genetic reduction algorithm and particle swarm optimization reduction has some of the global search ability and implicit parallelism, but genetic reduction algorithm does not have memory function, some better solutions may be with the change of population and is gradually destroyed, and particle swarm optimization reduction although can of particles in the population remained with memory function, can the better solution preserved, but particle swarm algorithm easy to fall into local optimal solution and sometimes difficult to convergence speed control. In order to overcome these problems to the greatest extent, this paper proposes an attribute reduction algorithm based on genetic algorithm and particle swarm optimization. the algorithm of the condition attribute decision attribute support degree based, the core attributes added to initialize the particle swarm to avoid randomly generated population blindness which can in some extent to accelerate the convergence of the algorithm, Reset adaptation degree function not only introduces the attribute dependent degree value as the basis of judgment and regulated the function parameters dynamically, in order to ensure the result is the minimum attributes reduction.And genetic algorithm combined with adaptive crossover and mutation operations, thus ensuring the particles feasible solution can be fully retained and utilized, make the algorithm in strengthening the ability of local search, while maintaining the global search ability. Experiments show that,especially in the large scale data set for minimum reduction, compared with other algorithms, has obvious advantages.