We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:

Download:

Current browse context:

math.ST

Change to browse by:

References & Citations

Bookmark

(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo ScienceWISE logo

Mathematics > Statistics Theory

Title: Leave-one-out Singular Subspace Perturbation Analysis for Spectral Clustering

Abstract: The singular subspaces perturbation theory is of fundamental importance in probability and statistics. It has various applications across different fields. We consider two arbitrary matrices where one is a leave-one-column-out submatrix of the other one and establish a novel perturbation upper bound for the distance between two corresponding singular subspaces. It is well-suited for mixture models and results in a sharper and finer statistical analysis than classical perturbation bounds such as Wedin's Theorem. Powered by this leave-one-out perturbation theory, we provide a deterministic entrywise analysis for the performance of the spectral clustering under mixture models. Our analysis leads to an explicit exponential error rate for the clustering of sub-Gaussian mixture models. For the mixture of isotropic Gaussians, the rate is optimal under a weaker signal-to-noise condition than that of L\"offler et al. (2021).
Subjects: Statistics Theory (math.ST); Machine Learning (cs.LG); Spectral Theory (math.SP)
Cite as: arXiv:2205.14855 [math.ST]
  (or arXiv:2205.14855v1 [math.ST] for this version)

Submission history

From: Anderson Ye Zhang [view email]
[v1] Mon, 30 May 2022 05:07:09 GMT (93kb)

Link back to: arXiv, form interface, contact.