机构地区: 广东轻工职业技术学院
出 处: 《现代计算机》 2008年第6期46-49,共4页
摘 要: 出于不同的应用目的,许多学者提出了各种不同的属性约简概念。给出相对熵保持不变的条件,利用相对熵定义了决策表的相对熵约简,证明相对熵约简与HU的差别矩阵和差别函数的约简方法是等价的。 For the sake of different applications, various definitions of attribute reduction have been proposed by different researchers. Gives the condition that keep relative entropy unchanged, and defines the relative entropy reduction of the decision tables through the relative entropy. Proves that the relative entropy reduction is equivalent to the HU's method of discernibility matrix and discernibility function.