Nonparametric Selection of the Smallest-dispersion Multivariate Population

Abstract

We present our recent attempts in selecting the smallest- dispersion multivariate population under a nonparametric setting. First we recall univariate nonparametric selection procedures, give motivations for this research, and review depth function for multivariate data. A depth based quantile is used to define the dispersion between populations. Then we propose selection procedures using the sample depth based quantiles. We are looking for applications in the areas other than signal processing and we welcome your suggestions.