基于Tabu搜索的粗糙集属性约简算法

Attribute Reduction Based on Tabu Search in Rough Set Theory

粗糙集理论中属性约简算法在保证解质量的情况下,效率比较低。针对这个问题提出一种基于记忆的启发式禁忌搜索算法,该算法称为TSAR(Tabu Search Attribute Reduction),是一个长期记忆的高性能TS算法。TSAR在利用邻域搜索方法的同时,又采用了广泛性和集中性模式,通过调用三个过程来产生及约简候选解,多参数智能化控制迭代次数,增大获得全局最优的机会,避免过早地陷入局部最优。TSAR和文献中算法相比,在解的质量上表现优异,而且计算的开销也很低。

In rough set theory, to keep solution qualities well, the efficiency of the attribute reduction is lower. In this paper, consider a memory- based heuristic of tabu search to solve the problem. The proposed method, called tabu search attribute reduction （TSAR）, is a high- level TS with long - term memory. Therefore, TSAR invokes diversification and intensification search schemes besides the TS neighborhood search methodology, by means of invoking three procedures to generate and reduct trial solutions, also intelligent controlling the iteration number with multi - parameter, to increase the opportunity of finding global optimal solution and to avoid falling into local optimal solution prematurely. In terms of solution qualities, TSAR shows promising and competitive performance compared with another algorithm in the listed document. Moreover, TSAR shows a superior performance in saving the computational costs.