一种基于进化算法的连续属性离散化方法

连续属性离散化是知识系统中的一个重要环节，一个好的离散化方法能够起到简化知识和描述和便于对知识系统的处理。而求取连续属性值的最优断点集合是一个NP难题，本文把连续属性值离散化问题作为一种约束优化问题，采用遗传算法来获得最优解，并针对离散化问题设计了相应的编码方式、交叉算子和变异算子。实验结果表明，采用遗传算法求解连续属性值最优断点集合是可行的。

A QUANTIZATION OF REAL-VALUE ATTRIBUTES BASED ON EVOLUTION ALGORITHM

The quantization of real-value attributes is an important process in the knowledge system.A good quantization can reduce the representation of knowledge and make easy to discovery knowledge.The optimal partition problem is a NP-hard.The quantization of real-value attributes is a constrained optimization and a new quantization based on evolution algorithm is proposed in this paper.To quantization this paper designs a code method,a specific crossover and mutation.The result of experiment indicates that the quantization based on evolution algorithm is efficient.