Efficient mining of discriminative co-clusters from gene expression data

摘要: Discriminative models are used to analyze the differences between two classes and identify class-specific patterns. Most of existing discriminative depend on using entire feature space compute patterns for each class. Co-clustering has been proposed capture that correlated in a subset features, but it cannot handle labeled datasets. In certain biological applications such as gene expression analysis, is critical consider only space. The objective this paper twofold: first, presents an algorithm efficiently find arbitrarily positioned co-clusters from complex data. Second, extends co-clustering discover by incorporating class information into co-cluster search process. addition, we also characterize propose three novel measures can be evaluate performance any subspace pattern-mining algorithm. We evaluated algorithms several synthetic real datasets, our experimental results showed outperformed available literature.