English

Minimax Optimal Rate for Parameter Estimation in Multivariate Deviated Models

Statistics Theory 2023-10-31 v2 Statistics Theory

Abstract

We study the maximum likelihood estimation (MLE) in the multivariate deviated model where the data are generated from the density function (1λ)h0(x)+λf(xμ,Σ)(1-\lambda^{\ast})h_{0}(x)+\lambda^{\ast}f(x|\mu^{\ast}, \Sigma^{\ast}) in which h0h_{0} is a known function, λ[0,1]\lambda^{\ast} \in [0,1] and (μ,Σ)(\mu^{\ast}, \Sigma^{\ast}) are unknown parameters to estimate. The main challenges in deriving the convergence rate of the MLE mainly come from two issues: (1) The interaction between the function h0h_{0} and the density function ff; (2) The deviated proportion λ\lambda^{\ast} can go to the extreme points of [0,1][0,1] as the sample size tends to infinity. To address these challenges, we develop the \emph{distinguishability condition} to capture the linear independent relation between the function h0h_{0} and the density function ff. We then provide comprehensive convergence rates of the MLE via the vanishing rate of λ\lambda^{\ast} to zero as well as the distinguishability of two functions h0h_{0} and ff.

Keywords

Cite

@article{arxiv.2301.11808,
  title  = {Minimax Optimal Rate for Parameter Estimation in Multivariate Deviated Models},
  author = {Dat Do and Huy Nguyen and Khai Nguyen and Nhat Ho},
  journal= {arXiv preprint arXiv:2301.11808},
  year   = {2023}
}

Comments

Dat Do and Huy Nguyen contributed equally to this work; 38 pages, 20 figures

R2 v1 2026-06-28T08:23:33.145Z