Random orthonormal embeddings are minimax optimal for sketched least squares, and rotation-invariant embeddings are minimax optimal for randomized SVD, with sharp error formulas.
Distributed sketching methods for privacy preserving regression
3 Pith papers cite this work. Polarity classification is still indexing.
3
Pith papers citing it
representative citing papers
NDIS lemma computes closed-form hockey-stick divergence δ(ε) between arbitrary multivariate Gaussians and is applied to obtain tighter privacy for random projection.
A multilevel sketch-and-solve estimator reduces variance faster than averaging independent sketches but incurs slightly higher computational cost, making the approach unattractive for least squares problems.
citing papers explorer
-
Sharp analysis of sketched least squares and randomized low-rank approximation
Random orthonormal embeddings are minimax optimal for sketched least squares, and rotation-invariant embeddings are minimax optimal for randomized SVD, with sharp error formulas.
-
The Normal Distributions Indistinguishability Spectrum and its Application to Privacy-Preserving Machine Learning
NDIS lemma computes closed-form hockey-stick divergence δ(ε) between arbitrary multivariate Gaussians and is applied to obtain tighter privacy for random projection.
-
A multilevel sketch-and-solve method for overdetermined least squares problems
A multilevel sketch-and-solve estimator reduces variance faster than averaging independent sketches but incurs slightly higher computational cost, making the approach unattractive for least squares problems.