现实中基于树型层次结构的属性值分类是普遍存在的,反映这种树型层次结构的属性值分类法(Attribute Value Taxonomy,AVT)已被证明对数据的泛化上是有效的。部分数据集已具备相关专家提供的AVT,但大多数数据集不具备人为提供的AVT。为此...现实中基于树型层次结构的属性值分类是普遍存在的,反映这种树型层次结构的属性值分类法(Attribute Value Taxonomy,AVT)已被证明对数据的泛化上是有效的。部分数据集已具备相关专家提供的AVT,但大多数数据集不具备人为提供的AVT。为此,本文提出一种VDM-AVT学习器,即一种依据数据自动构造AVT的方法;为了评价所构造AVT的质量,基于VDM-AVT学习器提出了VDM-AVT-AGR模型。VDM-AVT学习器基于VDM距离,利用层次聚类将属性值抽象为树型层次结构,VDM-AVT-AGR模型利用VDM-AVT学习器得到的AVT对数据进行属性泛化约简。实验表明,利用VDM-AVTAGRD模型处理后的数据集相比原始数据集具有更好的分类性能和泛化能力。由此也可以证明VDM-AVT学习器得出的AVT是有效的。展开更多
In his article 'On Information' published in the 1985 No.2 issue of SocialSciences in China Li Ming examines the process of information cognition of humanbeings and the history of the emergence of information ...In his article 'On Information' published in the 1985 No.2 issue of SocialSciences in China Li Ming examines the process of information cognition of humanbeings and the history of the emergence of information theory and furthermore,givesa definition of information derived from physical interactions.His observations arevery enlightening.However, there are still a few questions worth further study.First, in giving a definition to information,his first generalization is that in-formation is 'a universal attribute of matter.' And he further says,'Here,sub-展开更多
文摘现实中基于树型层次结构的属性值分类是普遍存在的,反映这种树型层次结构的属性值分类法(Attribute Value Taxonomy,AVT)已被证明对数据的泛化上是有效的。部分数据集已具备相关专家提供的AVT,但大多数数据集不具备人为提供的AVT。为此,本文提出一种VDM-AVT学习器,即一种依据数据自动构造AVT的方法;为了评价所构造AVT的质量,基于VDM-AVT学习器提出了VDM-AVT-AGR模型。VDM-AVT学习器基于VDM距离,利用层次聚类将属性值抽象为树型层次结构,VDM-AVT-AGR模型利用VDM-AVT学习器得到的AVT对数据进行属性泛化约简。实验表明,利用VDM-AVTAGRD模型处理后的数据集相比原始数据集具有更好的分类性能和泛化能力。由此也可以证明VDM-AVT学习器得出的AVT是有效的。
文摘In his article 'On Information' published in the 1985 No.2 issue of SocialSciences in China Li Ming examines the process of information cognition of humanbeings and the history of the emergence of information theory and furthermore,givesa definition of information derived from physical interactions.His observations arevery enlightening.However, there are still a few questions worth further study.First, in giving a definition to information,his first generalization is that in-formation is 'a universal attribute of matter.' And he further says,'Here,sub-
