Detection of clusters is a crucial task across many disciplines such as statistics, engineering and bioinformatics. We mainly focus on the modern high dimensional scenario, where traditional methods could fail due to the curse of dimensionality. In this study, we propose a non-parametric framework for clustering that can be applied to arbitrary dimensions. Simulation results show that this new framework outperforms the existing methods under a wide range of settings. We illustrate the proposed method on real data applications in distinguishing cancer tissues from normal tissues through gene expression data.
翻译:暂无翻译