English

Stepdown SLOPE for Controlled Feature Selection

Statistics Theory 2023-02-22 v1 Statistics Theory

Abstract

Sorted L-One Penalized Estimation (SLOPE) has shown the nice theoretical property as well as empirical behavior recently on the false discovery rate (FDR) control of high-dimensional feature selection by adaptively imposing the non-increasing sequence of tuning parameters on the sorted 1\ell_1 penalties. This paper goes beyond the previous concern limited to the FDR control by considering the stepdown-based SLOPE to control the probability of kk or more false rejections (kk-FWER) and the false discovery proportion (FDP). Two new SLOPEs, called kk-SLOPE and F-SLOPE, are proposed to realize kk-FWER and FDP control respectively, where the stepdown procedure is injected into the SLOPE scheme. For the proposed stepdown SLOPEs, we establish their theoretical guarantees on controlling kk-FWER and FDP under the orthogonal design setting, and also provide an intuitive guideline for the choice of regularization parameter sequence in much general setting. Empirical evaluations on simulated data validate the effectiveness of our approaches on controlled feature selection and support our theoretical findings.

Keywords

Cite

@article{arxiv.2302.10610,
  title  = {Stepdown SLOPE for Controlled Feature Selection},
  author = {Jingxuan Liang and Hong Chen and Xuelin Zhang and Weifu Li and Xin Tang},
  journal= {arXiv preprint arXiv:2302.10610},
  year   = {2023}
}