A Perturbation Bound on the Subspace Estimator from Canonical Projections

TL;DR

This paper establishes a perturbation bound for the subspace estimator derived from noisy canonical projections, impacting areas like matrix completion and subspace clustering.

Contribution

It provides a fundamental perturbation bound for subspace estimators based on canonical projections contaminated by noise.

Findings

Derived a new perturbation bound for subspace estimators

Implications for matrix completion and subspace clustering

Enhances understanding of noise effects on subspace estimation

Abstract

This paper derives a perturbation bound on the optimal subspace estimator obtained from a subset of its canonical projections contaminated by noise. This fundamental result has important implications in matrix completion, subspace clustering, and related problems.