mlsanityData to 7 Oct 2026

Rate-Optimal Perturbation Bounds for Singular Subspaces with Applications to High-Dimensional Statistics

arXiv 1605.00353 ↗First posted 2 May 2016math.ST math.PR stat.MEcited by 45

Not on the map. Why?

Abstract (arXiv)

Perturbation bounds for singular spaces, in particular Wedin's sin Theta theorem, are a fundamental tool in many fields including high-dimensional statistics, machine learning, and applied mathematics. In this paper, we establish separate perturbation bounds, measured in both spectral and Frobenius sin Theta distances, for the left and right singular subspaces. Lower bounds, which show that the individual perturbation bounds are rate-optimal, are also given. The new perturbation bounds are applicable to a wide range of problems. In this paper, we consider in detail applications to low-rank matrix denoising and singular space estimation, high-dimensional clustering, and canonical correlation analysis (CCA). In particular, separate matching upper and lower bounds are obtained for estimating the left and right singular spaces. To the best of our knowledge, this is the first result that gives different optimal rates for the left and right singular spaces under the same perturbation. In addition to these problems, applications to other high-dimensional problems such as community detection in bipartite networks, multidimensional scaling, and cross-covariance matrix estimation are also discussed.

Key facts. Cited by 45 papers in the map (data through 7 Oct 2026). First posted on arXiv 2 May 2016.

Builds on

No reference list yet.

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2021: 8; 2022: 9; 2023: 8; 2024: 5; 2025: 8; 2026: 7.

45 papers.

Sort:

By area:

  1. 2021Theoretical Foundations of t-SNE for Visualizing High-Dimensional Clustered Datacited by 55
  2. 2021The Power of Contrast for Feature Learning: A Theoretical Analysiscited by 53
  3. 2023Elliptic PDE learning is provably data-efficientcited by 26
  4. 2021Inference for Heteroskedastic PCA with Missing Datacited by 21
  5. 2021Privately Learning Subspacescited by 16
  6. 2022Nearly Minimax Algorithms for Linear Bandits with Shared Representationcited by 16
  7. 2022Tensor-on-Tensor Regression: Riemannian Optimization, Over-parameterization, Statistical-computational Gap, and Their Interplaycited by 15
  8. 2022Leave-one-out Singular Subspace Perturbation Analysis for Spectral Clusteringcited by 14
  9. 2021Latent Space Model for Higher-order Networks and Generalized Tensor Decompositioncited by 13
  10. 2022Regret Lower Bounds for Learning Linear Quadratic Gaussian Systemscited by 11
  11. 2023Approximate message passing from random initialization with applications to mathbbZ2 synchronizationcited by 9
  12. 2023Spectral Entry-wise Matrix Estimation for Low-Rank Reinforcement Learningcited by 9
33 more
  1. 2023Doubly High-Dimensional Contextual Bandits: An Interpretable Model for Joint Assortment-Pricingcited by 8
  2. 2021Learning Good State and Action Representations via Tensor Decompositioncited by 7
  3. 2023Deflated HeteroPCA: Overcoming the curse of ill-conditioning in heteroskedastic PCAcited by 7
  4. 2022Optimal Estimation and Computational Limit of Low-rank Gaussian Mixturescited by 6
  5. 2022A Spectral Method for Assessing and Combining Multiple Data Visualizationscited by 6
  6. 2024Optimal Estimation of Shared Singular Subspaces across Multiple Noisy Matricescited by 6
  7. 2022One-Way Matching of Datasets with Low Rank Signalscited by 5
  8. 2022Optimal Clustering by Lloyd Algorithm for Low-Rank Mixture Modelcited by 5
  9. 2024Learning Joint and Individual Structure in Network Data with Covariatescited by 2
  10. 2023Recovering Simultaneously Structured Data via Non-Convex Iteratively Reweighted Least Squarescited by 1
  11. 2023Mode-wise Principal Subspace Pursuit and Matrix Spiked Covariance Modelcited by 1
  12. 2024On Differentially Private Subspace Estimation in a Distribution-Free Settingcited by 1
  13. 2024Signed Diverse Multiplex Networks: Clustering and Inferencecited by 1
  14. 2025Provable Affine Identifiability of Nonlinear CCA under Latent Distributional Priorscited by 1
  15. 2025Concentration bounds on response-based vector embeddings of black-box generative modelscited by 1
  16. 2021Classification of high-dimensional data with spiked covariance matrix structurecited by 0
  17. 2021Recommending with Recommendationscited by 0
  18. 2022Entrywise Recovery Guarantees for Sparse PCA via Sparsistent Algorithmscited by 0
  19. 2023Statistical and computational rates in high rank tensor estimationcited by 0
  20. 2024On Minimum Trace Factor Analysis -- An Old Song Sung to a New Tunecited by 0
  21. 2025Subspace Recovery in Winsorized PCA: Insights into Accuracy and Robustnesscited by 0
  22. 2025High-dimensional Clustering and Signal Recovery under Block Signalscited by 0
  23. 2025Perfect Clustering in Very Sparse Diverse Multiplex Networkscited by 0
  24. 2025Optimal Bias-variance Tradeoff in Matrix and Tensor Estimationcited by 0
  25. 2025On Spectral Learning for Odeco Tensors: Perturbation, Initialization, and Algorithmscited by 0
  26. 2025Canonical correlation regression with noisy datacited by 0
  27. 2026SCaLE: Switching Cost aware Learning and Explorationcited by 0
  28. 2026Adjacency Spectral Embeddings of Correlation Networkscited by 0
  29. 2026Provable Subspace Identification of Nonlinear Multi-view CCAcited by 0
  30. 2026On the Spectral Structure and Objective Equivalence of Orthogonal Multilabel Fisher Discriminantscited by 0
  31. 2026Enhancing Spectral Embedding through Robust and Flexible Knowledge Transfer in Electronic Health Recordscited by 0
  32. 2026Enhancing Differentially Private Mechanisms via Empirical Bayescited by 0
  33. 2026When Does Dense Retrieval Need Asymmetric Geometry? A Bias-Variance Theory of Shared and Dual Projectionscited by 0