论文部分内容阅读
粗糙集理论已被证明是一种有效的属性约简方法.目前有许多启发式属性约简算法已被提出,其中基于信息熵的属性约简算法受到了广泛的关注.为此,针对现有的基于信息熵的属性约简算法问题,定义一种新的信息熵模型—–近似决策熵,并提出一种基于近似决策熵的属性约简(ADEAR)算法.通过在多个UCI数据集上的实验表明,与现有算法相比,ADEAR算法能够获得较小的约简和较高的分类精度,具有相对较低的计算开销.
Rough set theory has been proved to be an effective attribute reduction method.At present, many heuristic attribute reduction algorithms have been proposed, in which attribute reduction algorithm based on information entropy has received widespread attention.To this end, , An attribute-reduction algorithm based on information entropy is defined, a new information entropy model-approximate decision entropy is defined and an attribute reduction algorithm (ADEAR) based on approximate decision entropy is proposed. Experiments show that, compared with the existing algorithms, ADEAR algorithm can obtain a smaller reduction and higher classification accuracy, with a relatively low computational overhead.