Online covariance estimation improves inference for manifold optimization

Online Covariance Estimation for Averaged Stochastic Gradient Descent on Riemannian Manifolds

Information TheoryMachine Learning

Summary

Making accurate guesses about parameters on curved spaces (called manifolds) is hard because usual math operations don’t directly apply. The authors study a method to update these guesses step-by-step as more data arrives, while also estimating the uncertainty of these guesses online. They develop new ways to combine and adjust information while respecting the curved geometry, without needing complicated calculations or knowing when the process will stop. Their approach lets computers do reliable statistical analysis on these curved spaces while data streams in, demonstrated on a problem about finding patterns in data.

What this means in practice

Authors

Jingrong Guan, Liwei Jiang, Sen Na

Abstract

We study online inference for averaged Riemannian stochastic gradient descent (RSGD) when the unknown parameter lies on a manifold. Although the averaged RSGD iterate is asymptotically normal, practical inference requires estimating the unknown covariance operator in its limiting distribution. Existing online batch-means methods cannot be applied directly because a general Riemannian manifold lacks an addition operation, and the tangent space containing the Riemannian gradient varies across iterations. Through vector transport, we construct overlapping and non-overlapping covariance estimators that adaptively combine information from successive batches while accounting for the changing tangent spaces. The proposed estimators are online, require neither Riemannian Hessian computations nor a prespecified terminal sample size, and attain the same convergence rates as their Euclidean counterparts (Zhu et al., 2023). Together with asymptotic normality, these estimators enable online statistical inference for RSGD on manifolds. We illustrate the proposed framework through a detailed study of streaming principal component analysis on Grassmann manifolds.