Entangled Mean Estimation in High-Dimensions

TL;DR

This paper introduces an efficient algorithm for high-dimensional entangled mean estimation that nearly matches the theoretical error bounds, using iterative refinement and novel rejection sampling techniques.

Contribution

It presents a new computationally efficient algorithm with near-optimal error bounds for high-dimensional mean estimation in the subset-of-signals model.

Findings

Achieves near-optimal error bounds up to polylogarithmic factors.

Develops a rejection sampling method to filter noisy samples.

Introduces an iterative dimension-reduction strategy inspired by list-decodable learning.

Abstract

We study the task of high-dimensional entangled mean estimation in the subset-of-signals model. Specifically, given $N$ independent random points $x_1,\ldots,x_N$ in $\mathbb{R}^D$ and a parameter $\alpha \in (0, 1)$ such that each $x_i$ is drawn from a Gaussian with mean $\mu$ and unknown covariance, and an unknown $\alpha$-fraction of the points have identity-bounded covariances, the goal is to estimate the common mean $\mu$. The one-dimensional version of this task has received significant attention in theoretical computer science and statistics over the past decades. Recent work [LY20; CV24] has given near-optimal upper and lower bounds for the one-dimensional setting. On the other hand, our understanding of even the information-theoretic aspects of the multivariate setting has remained limited. In this work, we design a computationally efficient algorithm achieving an information-theoretically near-optimal error. Specifically, we show that the optimal error (up to polylogarithmic factors) is $f(\alpha,N) + \sqrt{D/(\alpha N)}$, where the term $f(\alpha,N)$ is the error of the one-dimensional problem and the second term is the sub-Gaussian error rate. Our algorithmic approach employs an iterative refinement strategy, whereby we progressively learn more accurate approximations $\hat \mu$ to $\mu$. This is achieved via a novel rejection sampling procedure that removes points significantly deviating from $\hat \mu$, as an attempt to filter out unusually noisy samples. A complication that arises is that rejection sampling introduces bias in the distribution of the remaining points. To address this issue, we perform a careful analysis of the bias, develop an iterative dimension-reduction strategy, and employ a novel subroutine inspired by list-decodable learning that leverages the one-dimensional result.