Optimal two-stage procedures for estimating location and size of the maximum of a multivariate regression function

TL;DR

This paper introduces a two-stage method for efficiently estimating the maximum's location and size of a multivariate regression function, improving convergence rates and partially overcoming the curse of dimensionality.

Contribution

It develops a two-stage procedure that achieves optimal convergence rates for estimating the maximum's location and size, even in high dimensions, under smoothness conditions.

Findings

Second stage estimators outperform first stage in convergence rates.

The method partially mitigates the curse of dimensionality.

Optimal rates are achieved for functions with sufficient smoothness.

Abstract

We propose a two-stage procedure for estimating the location $\bolds{\mu}$ and size M of the maximum of a smooth d-variate regression function f. In the first stage, a preliminary estimator of $\bolds{\mu}$ obtained from a standard nonparametric smoothing method is used. At the second stage, we "zoom-in" near the vicinity of the preliminary estimator and make further observations at some design points in that vicinity. We fit an appropriate polynomial regression model to estimate the location and size of the maximum. We establish that, under suitable smoothness conditions and appropriate choice of the zooming, the second stage estimators have better convergence rates than the corresponding first stage estimators of $\bolds{\mu}$ and M. More specifically, for $\alpha$-smooth regression functions, the optimal nonparametric rates $n^{-(\alpha-1)/(2\alpha+d)}$ and $n^{-\alpha/(2\alpha+d)}$ at the first stage can be improved to $n^{-(\alpha-1)/(2\alpha)}$ and $n^{-1/2}$, respectively, for $\alpha>1+\sqrt{1+d/2}$. These rates are optimal in the class of all possible sequential estimators. Interestingly, the two-stage procedure resolves "the curse of the dimensionality" problem to some extent, as the dimension d does not control the second stage convergence rates, provided that the function class is sufficiently smooth. We consider a multi-stage generalization of our procedure that attains the optimal rate for any smoothness level $\alpha>2$ starting with a preliminary estimator with any power-law rate at the first stage.