Matches in SemOpenAlex for { <https://semopenalex.org/work/W4386501855> ?p ?o ?g. }
Showing items 1 to 71 of
71
with 100 items per page.
- W4386501855 abstract "We study the differential privacy (DP) of a core ML problem, linear ordinary least squares (OLS), a.k.a. $ell_2$-regression. Our key result is that the approximate LS algorithm (ALS) (Sarlos, 2006), a randomized solution to the OLS problem primarily used to improve performance on large datasets, also preserves privacy. ALS achieves a better privacy/utility tradeoff, without modifications or further noising, when compared to alternative private OLS algorithms which modify and/or noise OLS. We give the first {em tight} DP-analysis for the ALS algorithm and the standard Gaussian mechanism (Dwork et al., 2014) applied to OLS. Our methodology directly improves the privacy analysis of (Blocki et al., 2012) and (Sheffet, 2019)) and introduces new tools which may be of independent interest: (1) the exact spectrum of $(epsilon, delta)$-DP parameters (``DP spectrum) for mechanisms whose output is a $d$-dimensional Gaussian, and (2) an improved DP spectrum for random projection (compared to (Blocki et al., 2012) and (Sheffet, 2019)). All methods for private OLS (including ours) assume, often implicitly, restrictions on the input database, such as bounds on leverage and residuals. We prove that such restrictions are necessary. Hence, computing the privacy of mechanisms such as ALS must estimate these database parameters, which can be infeasible in big datasets. For more complex ML models, DP bounds may not even be tractable. There is a need for blackbox DP-estimators (Lu et al., 2022) which empirically estimate a data-dependent privacy. We demonstrate the effectiveness of such a DP-estimator by empirically recovering a DP-spectrum that matches our theory for OLS. This validates the DP-estimator in a nontrivial ML application, opening the door to its use in more complex nonlinear ML settings where theory is unavailable." @default.
- W4386501855 created "2023-09-08" @default.
- W4386501855 creator A5014862413 @default.
- W4386501855 creator A5014988509 @default.
- W4386501855 creator A5026950074 @default.
- W4386501855 creator A5051863472 @default.
- W4386501855 date "2023-09-03" @default.
- W4386501855 modified "2023-10-16" @default.
- W4386501855 title "Privacy-Utility Tradeoff of OLS with Random Projections" @default.
- W4386501855 doi "https://doi.org/10.48550/arxiv.2309.01243" @default.
- W4386501855 hasPublicationYear "2023" @default.
- W4386501855 type Work @default.
- W4386501855 citedByCount "0" @default.
- W4386501855 crossrefType "posted-content" @default.
- W4386501855 hasAuthorship W4386501855A5014862413 @default.
- W4386501855 hasAuthorship W4386501855A5014988509 @default.
- W4386501855 hasAuthorship W4386501855A5026950074 @default.
- W4386501855 hasAuthorship W4386501855A5051863472 @default.
- W4386501855 hasBestOaLocation W43865018551 @default.
- W4386501855 hasConcept C105795698 @default.
- W4386501855 hasConcept C108827166 @default.
- W4386501855 hasConcept C11413529 @default.
- W4386501855 hasConcept C119857082 @default.
- W4386501855 hasConcept C121332964 @default.
- W4386501855 hasConcept C123201435 @default.
- W4386501855 hasConcept C126255220 @default.
- W4386501855 hasConcept C153083717 @default.
- W4386501855 hasConcept C154945302 @default.
- W4386501855 hasConcept C163716315 @default.
- W4386501855 hasConcept C185429906 @default.
- W4386501855 hasConcept C23130292 @default.
- W4386501855 hasConcept C2776441110 @default.
- W4386501855 hasConcept C2777036070 @default.
- W4386501855 hasConcept C33923547 @default.
- W4386501855 hasConcept C41008148 @default.
- W4386501855 hasConcept C62520636 @default.
- W4386501855 hasConcept C99656134 @default.
- W4386501855 hasConceptScore W4386501855C105795698 @default.
- W4386501855 hasConceptScore W4386501855C108827166 @default.
- W4386501855 hasConceptScore W4386501855C11413529 @default.
- W4386501855 hasConceptScore W4386501855C119857082 @default.
- W4386501855 hasConceptScore W4386501855C121332964 @default.
- W4386501855 hasConceptScore W4386501855C123201435 @default.
- W4386501855 hasConceptScore W4386501855C126255220 @default.
- W4386501855 hasConceptScore W4386501855C153083717 @default.
- W4386501855 hasConceptScore W4386501855C154945302 @default.
- W4386501855 hasConceptScore W4386501855C163716315 @default.
- W4386501855 hasConceptScore W4386501855C185429906 @default.
- W4386501855 hasConceptScore W4386501855C23130292 @default.
- W4386501855 hasConceptScore W4386501855C2776441110 @default.
- W4386501855 hasConceptScore W4386501855C2777036070 @default.
- W4386501855 hasConceptScore W4386501855C33923547 @default.
- W4386501855 hasConceptScore W4386501855C41008148 @default.
- W4386501855 hasConceptScore W4386501855C62520636 @default.
- W4386501855 hasConceptScore W4386501855C99656134 @default.
- W4386501855 hasLocation W43865018551 @default.
- W4386501855 hasOpenAccess W4386501855 @default.
- W4386501855 hasPrimaryLocation W43865018551 @default.
- W4386501855 hasRelatedWork W1989839813 @default.
- W4386501855 hasRelatedWork W1990534671 @default.
- W4386501855 hasRelatedWork W2515196032 @default.
- W4386501855 hasRelatedWork W2937255903 @default.
- W4386501855 hasRelatedWork W2953239333 @default.
- W4386501855 hasRelatedWork W2953892398 @default.
- W4386501855 hasRelatedWork W3033759733 @default.
- W4386501855 hasRelatedWork W4292871255 @default.
- W4386501855 hasRelatedWork W4321612618 @default.
- W4386501855 hasRelatedWork W2185539502 @default.
- W4386501855 isParatext "false" @default.
- W4386501855 isRetracted "false" @default.
- W4386501855 workType "article" @default.