论文部分内容阅读
针对传统生物启发式方法在决策表中属性约简求解效率不高和难以协同约简等问题,提出一种基于量子混合协同进化的自适应多级联属性约简算法. 首先设计了一种新型高效的自适应量子角旋转策略,指导参与属性约简的进化种群自适应相互演进,加速算法收敛. 然后构建了合作和竞争混合的协同进化级联模型,根据执行经验记录分割属性种群集,提高约简子种群的多样性,并产生种群精英以增强其寻优经验共享,快速找到全局最小属性约简集. 实验结果表明,与同类典型算法相比,该算法在最小属性约简效率和精度方面具有明显优势.
Aiming at the problems of traditional bio-heuristic methods, such as the inefficiency of attribute reduction in decision table and the difficulty of collaborative reduction, a new adaptive multi-cascade attribute reduction algorithm based on quantum hybrid co-evolution is proposed. Efficient adaptive quantum rotation strategy to guide evolutionary population adaptive reduction of attribute reduction to accelerate each other and accelerate the convergence of the algorithm.Then a co-evolutionary cascade model of cooperative and competitive hybrid is constructed and the attribute population is segmented according to the execution experience And simplifies the population diversity and produces the elite population to enhance their sharing experience and quickly find the global minimal attribute reduction set.Experimental results show that compared with the typical algorithms, this algorithm has the advantages of minimum attribute reduction efficiency and precision Has obvious advantages.