Cluster-Seeking James-Stein Estimators
This addresses the challenge of designing effective estimators without prior parameter information, offering improvements in statistical estimation for high-dimensional data analysis.
The paper tackles the problem of estimating high-dimensional parameters from noisy observations by proposing shrinkage estimators that cluster data components and shrink them towards inferred attractors, achieving significant risk reduction over the maximum-likelihood estimator for a wide range of parameters, especially in large dimensions.
This paper considers the problem of estimating a high-dimensional vector of parameters $\boldsymbolθ \in \mathbb{R}^n$ from a noisy observation. The noise vector is i.i.d. Gaussian with known variance. For a squared-error loss function, the James-Stein (JS) estimator is known to dominate the simple maximum-likelihood (ML) estimator when the dimension $n$ exceeds two. The JS-estimator shrinks the observed vector towards the origin, and the risk reduction over the ML-estimator is greatest for $\boldsymbolθ$ that lie close to the origin. JS-estimators can be generalized to shrink the data towards any target subspace. Such estimators also dominate the ML-estimator, but the risk reduction is significant only when $\boldsymbolθ$ lies close to the subspace. This leads to the question: in the absence of prior information about $\boldsymbolθ$, how do we design estimators that give significant risk reduction over the ML-estimator for a wide range of $\boldsymbolθ$? In this paper, we propose shrinkage estimators that attempt to infer the structure of $\boldsymbolθ$ from the observed data in order to construct a good attracting subspace. In particular, the components of the observed vector are separated into clusters, and the elements in each cluster shrunk towards a common attractor. The number of clusters and the attractor for each cluster are determined from the observed vector. We provide concentration results for the squared-error loss and convergence results for the risk of the proposed estimators. The results show that the estimators give significant risk reduction over the ML-estimator for a wide range of $\boldsymbolθ$, particularly for large $n$. Simulation results are provided to support the theoretical claims.