Inference with non-differentiable surrogate loss in a general high-dimensional classification framework

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Liang, Muxuan, Ning, Yang, Smith, Maureen A, Zhao, Ying-Qi
Natura: Preprint
Pubblicazione: 2024
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866918481715265536
author Liang, Muxuan
Ning, Yang
Smith, Maureen A
Zhao, Ying-Qi
author_facet Liang, Muxuan
Ning, Yang
Smith, Maureen A
Zhao, Ying-Qi
contents Penalized empirical risk minimization with a surrogate loss function is often used to learn a high-dimensional linear decision rule in classification problems. Although much of the literature focus on the generalization error, there is a lack of inference procedures for identifying the driving factors of the estimated decision rule, especially when the surrogate loss is non-differentiable. We propose a kernel-smoothed decorrelated score to construct hypothesis tests and interval estimators for a linear decision rule estimated using a piece-wise linear surrogate loss, which has a discontinuous gradient and non-regular Hessian. Specifically, we adopt kernel approximations to smooth the discontinuous gradient near discontinuity points and approximate the non-regular Hessian of the surrogate loss. In applications where additional nuisance parameters are involved, we propose a novel cross-fitted version to accommodate flexible nuisance estimates and kernel approximations. We establish the limiting distribution of the kernel-smoothed decorrelated score and its cross-fitted version in a high-dimensional setup. Simulation and real data analysis are conducted to demonstrate the validity and the superiority of the proposed method.
format Preprint
id arxiv_https___arxiv_org_abs_2405_11723
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Inference with non-differentiable surrogate loss in a general high-dimensional classification framework
Liang, Muxuan
Ning, Yang
Smith, Maureen A
Zhao, Ying-Qi
Methodology
Machine Learning
Penalized empirical risk minimization with a surrogate loss function is often used to learn a high-dimensional linear decision rule in classification problems. Although much of the literature focus on the generalization error, there is a lack of inference procedures for identifying the driving factors of the estimated decision rule, especially when the surrogate loss is non-differentiable. We propose a kernel-smoothed decorrelated score to construct hypothesis tests and interval estimators for a linear decision rule estimated using a piece-wise linear surrogate loss, which has a discontinuous gradient and non-regular Hessian. Specifically, we adopt kernel approximations to smooth the discontinuous gradient near discontinuity points and approximate the non-regular Hessian of the surrogate loss. In applications where additional nuisance parameters are involved, we propose a novel cross-fitted version to accommodate flexible nuisance estimates and kernel approximations. We establish the limiting distribution of the kernel-smoothed decorrelated score and its cross-fitted version in a high-dimensional setup. Simulation and real data analysis are conducted to demonstrate the validity and the superiority of the proposed method.
title Inference with non-differentiable surrogate loss in a general high-dimensional classification framework
topic Methodology
Machine Learning
url https://arxiv.org/abs/2405.11723