Papers › PCA from noisy, linearly reduced data: the diagonal case
PCA from noisy, linearly reduced data: the diagonal case
Edgar Dobriban, William Leeb, Amit Singer
The archive published only this paper's code-link row. Authors, date and abstract are from arXiv's metadata (CC0), read from the Kaggle arXiv metadata snapshot of 2026-09-12 where its title matched the archive's; the title is the archive's.
Suppose we observe data of the form Yᵢ = Dᵢ (Sᵢ + εᵢ) ∈ℝᵖ or Yᵢ = Dᵢ Sᵢ + εᵢ ∈ℝᵖ, i=1,…,n, where Dᵢ ∈ℝ^(p×p) are known diagonal matrices, εᵢ are noise, and we wish to perform principal component analysis (PCA) on the unobserved signals Sᵢ ∈ℝᵖ. The first model arises in missing data problems, where the Dᵢ are binary. The second model captures noisy deconvolution problems, where the Dᵢ are the Fourier transforms of the convolution kernels. It is often reasonable to assume the Sᵢ lie on an unknown low-dimensional linear space; however, because many coordinates can be suppressed by the Dᵢ, this low-dimensional structure can be obscured. We introduce diagonally reduced spiked covariance models to capture this setting. We characterize the behavior of the singular vectors and singular values of the data matrix under high-dimensional asymptotics where n,p→∞ such that p/n→γ>0. Our results have the most general assumptions to date even without diagonal reduction. Using them, we develop optimal eigenvalue shrinkage methods for covariance matrix estimation and optimal singular value shrinkage methods for data denoising. Finally, we characterize the error rates of the empirical Best Linear Predictor (EBLP) denoisers. We show that, perhaps surprisingly, their optimal tuning depends on whether we denoise in-sample or out-of-sample, but the optimally tuned mean squared error is the same in the two cases.
Code
Repository list and official/mentioned flags are the archive's, frozen 2025-07-28. Reachability, where shown, is from one Syntology probe window (2026-09-16 to 2026-09-18); repositories not probed show nothing. GitHub stars are not tracked.
Code Syntology ran Syntology
Not run by Syntology. Nothing on this page verifies that the listed code works.
Results from the paper archive 2025-07-28
No leaderboard rows for this paper in the archive.
Report a problem or propose a change · a person checks every report against the paper or source before anything changes; decisions are listed on /corrections