非共现数据两阶段加权IB算法

非共现数据是指不符合联合概率分布,而是符合一个未知函数的数据.将非共现数据转化为共现形式后可以采用熵来定量度量信息并进行聚类.但是,现有算法假设非共现数据的各个属性特征对聚类贡献均匀,没有考虑代表性属性和不相关(冗余)属性对聚类效果的不同影响.因此,本文提出一个非共现数据的两阶段加权IB算法(TSAW-sIB),在非共现数据共现转化的两个阶段,从"非共现/共现/联合"三个视角观察非共现数据,突出代表性属性,抑制冗余属性,获得更能准确反映非共现数据特征的数据表示并进行聚类.实验表明,TSAW-sIB算法优于ROCK、COOLCAT和LIMBO算法.

Two-stage Attribute Weighting IB Algorithm for Non Co-occurrence Data

Non co-occurrence data does not appear in the form of co-occurrence of two variables X,Y,but rather as a sample of values of an unknown function Z(X,Y).The co-occurrence transformation of non co-occurrence data is necessary for clustering on the concept of Shannon entropy.However,these clustering algorithms treat all features fairly and set weights of all features equally.Therefore,the paper proposes a two-stage attribute weighting IB Algorithm(TPAW-sIB).At two stages of the co-occurrence transformation,we highlight representative features and dim irrelevant features from three viewpoints:non co-occurrence,co-occurrence and both.Experiments show that the TPAW-sIB algorithm is superior to the ROCK algorithm,the COOLCAT algorithm and the LIMBO algorithm.