FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W3099692016> ?p ?o ?g. }

• W3099692016 abstract "Balancing the privacy-utility tradeoff is a crucial requirement of many practical machine learning systems that deal with sensitive customer data. A popular approach for privacy- preserving text analysis is noise injection, in which text data is first mapped into a continuous embedding space, perturbed by sampling a spherical noise from an appropriate distribution, and then projected back to the discrete vocabulary space. While this allows the perturbation to admit the required metric differential privacy, often the utility of downstream tasks modeled on this perturbed data is low because the spherical noise does not account for the variability in the density around different words in the embedding space. In particular, words in a sparse region are likely unchanged even when the noise scale is large. In this paper, we propose a text perturbation mechanism based on a carefully designed regularized variant of the Mahalanobis metric to overcome this problem. For any given noise scale, this metric adds an elliptical noise to account for the covariance structure in the embedding space. This heterogeneity in the noise scale along different directions helps ensure that the words in the sparse region have sufficient likelihood of replacement without sacrificing the overall utility. We provide a text-perturbation algorithm based on this metric and formally prove its privacy guarantees. Additionally, we empirically show that our mechanism improves the privacy statistics to achieve the same level of utility as compared to the state-of-the-art Laplace mechanism." @default.
• W3099692016 created "2020-11-23" @default.
• W3099692016 creator A5021612761 @default.
• W3099692016 creator A5039434142 @default.
• W3099692016 creator A5068227018 @default.
• W3099692016 creator A5070387976 @default.
• W3099692016 date "2020-01-01" @default.
• W3099692016 modified "2023-10-11" @default.
• W3099692016 title "A Differentially Private Text Perturbation Method Using Regularized Mahalanobis Metric" @default.
• W3099692016 cites W145956381 @default.
• W3099692016 cites W1540596182 @default.
• W3099692016 cites W1658920975 @default.
• W3099692016 cites W1690606251 @default.
• W3099692016 cites W1711940457 @default.
• W3099692016 cites W1873763122 @default.
• W3099692016 cites W1904610397 @default.
• W3099692016 cites W1975937116 @default.
• W3099692016 cites W1989021088 @default.
• W3099692016 cites W1990512452 @default.
• W3099692016 cites W1996118086 @default.
• W3099692016 cites W2026002400 @default.
• W3099692016 cites W2027595342 @default.
• W3099692016 cites W2040228409 @default.
• W3099692016 cites W2056700697 @default.
• W3099692016 cites W2062112832 @default.
• W3099692016 cites W2082894754 @default.
• W3099692016 cites W2095272373 @default.
• W3099692016 cites W2106139345 @default.
• W3099692016 cites W2109426455 @default.
• W3099692016 cites W2115972246 @default.
• W3099692016 cites W2116950384 @default.
• W3099692016 cites W2124090777 @default.
• W3099692016 cites W2160495867 @default.
• W3099692016 cites W2181631377 @default.
• W3099692016 cites W2188999827 @default.
• W3099692016 cites W2198253679 @default.
• W3099692016 cites W2245160765 @default.
• W3099692016 cites W2250539671 @default.
• W3099692016 cites W2288752886 @default.
• W3099692016 cites W2476457629 @default.
• W3099692016 cites W2493916176 @default.
• W3099692016 cites W2535690855 @default.
• W3099692016 cites W2901147448 @default.
• W3099692016 cites W2955363520 @default.
• W3099692016 cites W2963378725 @default.
• W3099692016 cites W2963626623 @default.
• W3099692016 cites W2998378988 @default.
• W3099692016 cites W3003815046 @default.
• W3099692016 cites W3123972088 @default.
• W3099692016 cites W331889179 @default.
• W3099692016 doi "https://doi.org/10.18653/v1/2020.privatenlp-1.2" @default.
• W3099692016 hasPublicationYear "2020" @default.
• W3099692016 type Work @default.
• W3099692016 sameAs 3099692016 @default.
• W3099692016 citedByCount "5" @default.
• W3099692016 countsByYear W30996920162021 @default.
• W3099692016 countsByYear W30996920162022 @default.
• W3099692016 crossrefType "proceedings-article" @default.
• W3099692016 hasAuthorship W3099692016A5021612761 @default.
• W3099692016 hasAuthorship W3099692016A5039434142 @default.
• W3099692016 hasAuthorship W3099692016A5068227018 @default.
• W3099692016 hasAuthorship W3099692016A5070387976 @default.
• W3099692016 hasBestOaLocation W30996920161 @default.
• W3099692016 hasConcept C105795698 @default.
• W3099692016 hasConcept C11413529 @default.
• W3099692016 hasConcept C115961682 @default.
• W3099692016 hasConcept C126255220 @default.
• W3099692016 hasConcept C154945302 @default.
• W3099692016 hasConcept C178650346 @default.
• W3099692016 hasConcept C1921717 @default.
• W3099692016 hasConcept C23130292 @default.
• W3099692016 hasConcept C33923547 @default.
• W3099692016 hasConcept C41008148 @default.
• W3099692016 hasConcept C41608201 @default.
• W3099692016 hasConcept C4199805 @default.
• W3099692016 hasConcept C80444323 @default.
• W3099692016 hasConcept C99498987 @default.
• W3099692016 hasConceptScore W3099692016C105795698 @default.
• W3099692016 hasConceptScore W3099692016C11413529 @default.
• W3099692016 hasConceptScore W3099692016C115961682 @default.
• W3099692016 hasConceptScore W3099692016C126255220 @default.
• W3099692016 hasConceptScore W3099692016C154945302 @default.
• W3099692016 hasConceptScore W3099692016C178650346 @default.
• W3099692016 hasConceptScore W3099692016C1921717 @default.
• W3099692016 hasConceptScore W3099692016C23130292 @default.
• W3099692016 hasConceptScore W3099692016C33923547 @default.
• W3099692016 hasConceptScore W3099692016C41008148 @default.
• W3099692016 hasConceptScore W3099692016C41608201 @default.
• W3099692016 hasConceptScore W3099692016C4199805 @default.
• W3099692016 hasConceptScore W3099692016C80444323 @default.
• W3099692016 hasConceptScore W3099692016C99498987 @default.
• W3099692016 hasLocation W30996920161 @default.
• W3099692016 hasLocation W30996920162 @default.
• W3099692016 hasOpenAccess W3099692016 @default.
• W3099692016 hasPrimaryLocation W30996920161 @default.
• W3099692016 hasRelatedWork W1980507268 @default.
• W3099692016 hasRelatedWork W1991269640 @default.
• W3099692016 hasRelatedWork W2006158573 @default.
• W3099692016 hasRelatedWork W2033000528 @default.
• W3099692016 hasRelatedWork W2054182835 @default.

• next