Saved in:
Bibliographic Details
Main Authors: Chen, Xin, Klusowski, Jason M.
Format: Preprint
Published: 2024
Subjects:
Online Access:https://arxiv.org/abs/2401.00691
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866918265732726784
author Chen, Xin
Klusowski, Jason M.
author_facet Chen, Xin
Klusowski, Jason M.
contents This paper introduces an iterative algorithm for training nonparametric additive models that enjoys favorable memory storage and computational requirements. The algorithm can be viewed as the functional counterpart of stochastic gradient descent, applied to the coefficients of a truncated basis expansion of the component functions. We show that the resulting estimator satisfies an oracle inequality that allows for model mis-specification. In the well-specified setting, by choosing the learning rate carefully across three distinct stages of training, we demonstrate that its risk is minimax optimal in terms of the dependence on both the dimensionality of the data and the size of the training sample. Unlike past work, we also provide polynomial convergence rates even when the covariates do not have full support on their domain.
format Preprint
id arxiv_https___arxiv_org_abs_2401_00691
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Stochastic Gradient Descent for Nonparametric Additive Regression
Chen, Xin
Klusowski, Jason M.
Machine Learning
This paper introduces an iterative algorithm for training nonparametric additive models that enjoys favorable memory storage and computational requirements. The algorithm can be viewed as the functional counterpart of stochastic gradient descent, applied to the coefficients of a truncated basis expansion of the component functions. We show that the resulting estimator satisfies an oracle inequality that allows for model mis-specification. In the well-specified setting, by choosing the learning rate carefully across three distinct stages of training, we demonstrate that its risk is minimax optimal in terms of the dependence on both the dimensionality of the data and the size of the training sample. Unlike past work, we also provide polynomial convergence rates even when the covariates do not have full support on their domain.
title Stochastic Gradient Descent for Nonparametric Additive Regression
topic Machine Learning
url https://arxiv.org/abs/2401.00691