LGDSSTMLOct 24, 2023

Online Robust Mean Estimation

arXiv:2310.15932v15 citationsh-index: 48
Originality Incremental advance
AI Analysis

This addresses the challenge of real-time robust data aggregation in systems like sensor networks, offering incremental improvements over offline approaches.

The paper tackles the problem of high-dimensional robust mean estimation in an online setting where sensors report readings over time, with some fraction behaving maliciously, and it provides an efficient online algorithm achieving an error bound of O(δ log(T)), nearly competitive with offline methods, and shows inefficient algorithms can achieve error independent of T under stronger assumptions.

We study the problem of high-dimensional robust mean estimation in an online setting. Specifically, we consider a scenario where $n$ sensors are measuring some common, ongoing phenomenon. At each time step $t=1,2,\ldots,T$, the $i^{th}$ sensor reports its readings $x^{(i)}_t$ for that time step. The algorithm must then commit to its estimate $μ_t$ for the true mean value of the process at time $t$. We assume that most of the sensors observe independent samples from some common distribution $X$, but an $ε$-fraction of them may instead behave maliciously. The algorithm wishes to compute a good approximation $μ$ to the true mean $μ^\ast := \mathbf{E}[X]$. We note that if the algorithm is allowed to wait until time $T$ to report its estimate, this reduces to the well-studied problem of robust mean estimation. However, the requirement that our algorithm produces partial estimates as the data is coming in substantially complicates the situation. We prove two main results about online robust mean estimation in this model. First, if the uncorrupted samples satisfy the standard condition of $(ε,δ)$-stability, we give an efficient online algorithm that outputs estimates $μ_t$, $t \in [T],$ such that with high probability it holds that $\|μ-μ^\ast\|_2 = O(δ\log(T))$, where $μ= (μ_t)_{t \in [T]}$. We note that this error bound is nearly competitive with the best offline algorithms, which would achieve $\ell_2$-error of $O(δ)$. Our second main result shows that with additional assumptions on the input (most notably that $X$ is a product distribution) there are inefficient algorithms whose error does not depend on $T$ at all.

Foundations

The foundational work for this paper's niche, ranked by how specifically the neighbourhood builds on it — not by global fame.

Your Notes