Matches in SemOpenAlex for { <https://semopenalex.org/work/W3164243831> ?p ?o ?g. }
- W3164243831 abstract "Ensuring strong theoretical privacy guarantees on text data is a challenging problem which is usually attained at the expense of utility. However, to improve the practicality of privacy preserving text analyses, it is essential to design algorithms that better optimize this tradeoff. To address this challenge, we propose a release mechanism that takes any (text) embedding vector as input and releases a corresponding private vector. The mechanism satisfies an extension of differential privacy to metric spaces. Our idea based on first randomly projecting the vectors to a lower-dimensional space and then adding noise in this projected space generates private vectors that achieve strong theoretical guarantees on its utility. We support our theoretical proofs with empirical experiments on multiple word embedding models and NLP datasets, achieving in some cases more than 10% gains over the existing state-of-the-art privatization techniques." @default.
- W3164243831 created "2021-06-07" @default.
- W3164243831 creator A5021612761 @default.
- W3164243831 creator A5036801391 @default.
- W3164243831 date "2021-01-01" @default.
- W3164243831 modified "2023-09-23" @default.
- W3164243831 title "Private Release of Text Embedding Vectors" @default.
- W3164243831 cites W1500518989 @default.
- W3164243831 cites W1602085912 @default.
- W3164243831 cites W1644055776 @default.
- W3164243831 cites W1658920975 @default.
- W3164243831 cites W1711940457 @default.
- W3164243831 cites W1873763122 @default.
- W3164243831 cites W1878405293 @default.
- W3164243831 cites W1937834619 @default.
- W3164243831 cites W1992926795 @default.
- W3164243831 cites W2014902591 @default.
- W3164243831 cites W2027595342 @default.
- W3164243831 cites W2028584480 @default.
- W3164243831 cites W2070246124 @default.
- W3164243831 cites W2082894754 @default.
- W3164243831 cites W2103142290 @default.
- W3164243831 cites W2119874464 @default.
- W3164243831 cites W2153579005 @default.
- W3164243831 cites W2159024459 @default.
- W3164243831 cites W2160660844 @default.
- W3164243831 cites W2163455955 @default.
- W3164243831 cites W2170540710 @default.
- W3164243831 cites W2245160765 @default.
- W3164243831 cites W2250539671 @default.
- W3164243831 cites W2251939518 @default.
- W3164243831 cites W2470101972 @default.
- W3164243831 cites W2493916176 @default.
- W3164243831 cites W2557283755 @default.
- W3164243831 cites W2789226849 @default.
- W3164243831 cites W2901147448 @default.
- W3164243831 cites W2946930197 @default.
- W3164243831 cites W2952604841 @default.
- W3164243831 cites W2963502357 @default.
- W3164243831 cites W2963918774 @default.
- W3164243831 cites W2963999993 @default.
- W3164243831 cites W2964165804 @default.
- W3164243831 cites W2970172244 @default.
- W3164243831 cites W2997422449 @default.
- W3164243831 cites W2998378988 @default.
- W3164243831 cites W3003815046 @default.
- W3164243831 cites W3038533766 @default.
- W3164243831 cites W3045013345 @default.
- W3164243831 cites W3082707687 @default.
- W3164243831 cites W3096738375 @default.
- W3164243831 cites W3099692016 @default.
- W3164243831 cites W3102407811 @default.
- W3164243831 cites W74946117 @default.
- W3164243831 doi "https://doi.org/10.18653/v1/2021.trustnlp-1.3" @default.
- W3164243831 hasPublicationYear "2021" @default.
- W3164243831 type Work @default.
- W3164243831 sameAs 3164243831 @default.
- W3164243831 citedByCount "1" @default.
- W3164243831 countsByYear W31642438312022 @default.
- W3164243831 crossrefType "proceedings-article" @default.
- W3164243831 hasAuthorship W3164243831A5021612761 @default.
- W3164243831 hasAuthorship W3164243831A5036801391 @default.
- W3164243831 hasBestOaLocation W31642438311 @default.
- W3164243831 hasConcept C108710211 @default.
- W3164243831 hasConcept C111472728 @default.
- W3164243831 hasConcept C111919701 @default.
- W3164243831 hasConcept C11413529 @default.
- W3164243831 hasConcept C115961682 @default.
- W3164243831 hasConcept C118615104 @default.
- W3164243831 hasConcept C124101348 @default.
- W3164243831 hasConcept C13336665 @default.
- W3164243831 hasConcept C138885662 @default.
- W3164243831 hasConcept C154945302 @default.
- W3164243831 hasConcept C162324750 @default.
- W3164243831 hasConcept C176217482 @default.
- W3164243831 hasConcept C198043062 @default.
- W3164243831 hasConcept C199360897 @default.
- W3164243831 hasConcept C202444582 @default.
- W3164243831 hasConcept C21547014 @default.
- W3164243831 hasConcept C23130292 @default.
- W3164243831 hasConcept C2524010 @default.
- W3164243831 hasConcept C2777462759 @default.
- W3164243831 hasConcept C2778029271 @default.
- W3164243831 hasConcept C2778572836 @default.
- W3164243831 hasConcept C33923547 @default.
- W3164243831 hasConcept C41008148 @default.
- W3164243831 hasConcept C41608201 @default.
- W3164243831 hasConcept C80444323 @default.
- W3164243831 hasConcept C89611455 @default.
- W3164243831 hasConcept C90805587 @default.
- W3164243831 hasConcept C99498987 @default.
- W3164243831 hasConceptScore W3164243831C108710211 @default.
- W3164243831 hasConceptScore W3164243831C111472728 @default.
- W3164243831 hasConceptScore W3164243831C111919701 @default.
- W3164243831 hasConceptScore W3164243831C11413529 @default.
- W3164243831 hasConceptScore W3164243831C115961682 @default.
- W3164243831 hasConceptScore W3164243831C118615104 @default.
- W3164243831 hasConceptScore W3164243831C124101348 @default.
- W3164243831 hasConceptScore W3164243831C13336665 @default.
- W3164243831 hasConceptScore W3164243831C138885662 @default.