Inference in high-dimensional single-index models under symmetric designs

Hamid Eftekhari, Moulinath Banerjee, Ya'acov Ritov

Research output: Contribution to journalArticlepeer-review

8 Scopus citations

Abstract

The problem of statistical inference for regression coefficients in a high-dimensional singleindex model is considered. Under elliptical symmetry, the single index model can be reformulated as a proxy linear model whose regression parameter is identifiable. We construct estimates of the regression coefficients of interest that are similar to the debiased lasso estimates in the standard linear model and exhibit similar properties: P n-consistency and asymptotic normality. The procedure completely bypasses the estimation of the unknown link function, which can be extremely challenging depending on the underlying structure of the problem. Furthermore, under Gaussianity, we propose more efficient estimates of the coefficients by expanding the link function in the Hermite polynomial basis. Finally, we illustrate our approach via carefully designed simulation experiments.

Original languageEnglish
JournalJournal of Machine Learning Research
Volume22
StatePublished - 2021
Externally publishedYes

Bibliographical note

Publisher Copyright:
© 2021 Microtome Publishing. All rights reserved.

Keywords

  • Compressed sensing
  • Debiased inference
  • Hermite polynomials
  • Sparsity

Fingerprint

Dive into the research topics of 'Inference in high-dimensional single-index models under symmetric designs'. Together they form a unique fingerprint.

Cite this