一种改进的Rough集属性约简启发式遗传算法

属性约简是知识发现中的关键问题之一 .为了能够有效地获取决策表中属性最小相对约简 ,提出了一种在优化初始群体基础上提高算法性能的启发式遗传算法 .首先 ,通过构造一个新的算子 ,将信息论角度定义的属性重要性度量作为启发式信息 ,来描述所选择的属性子集对论域中确定分类子集的影响 ;接着 ,以此为基础并结合遗传算法 ,选择一些经过优化的染色体作为初始群体 ,在加强局部搜索能力的同时保持了该算法全局寻优的特性 .最后 ,从理论上对算法做了分析 ,证明了新算子所选择的属性子集对原有属性分类能力保持不变 .试验分析表明 ,该算法能有效地对决策表属性进行约简

A modified heuristic genetic algorithm for reduction of attributes in rough set theory

Attribute reduction is one of the key problems in knowledge discovery. A modified heuristic genetic algorithm based on optimizing initial population is proposed to effectively achieve the minimal relative reduction of the attributes in a decision table. The effect of attribute subclass on certain classification subset in universe is described by constructing a new operator and regarding the significance of the attributes defined from the viewpoint of information theory as heuristic information. Then, the optimized chromosomes are selected as initial population in order to enhance the ability of local search of the algorithm and to maintain the feature of global search of it. Finally, the algorithm is analyzed in theory and it is proven that its local search ability is enhanced and its global search specialty is maintained simultaneously. The experimental results show that this algorithm is effective to the attribute reduction of decision tables.