자료유형 | 학위논문 |
---|---|
서명/저자사항 | Applying Differential Privacy with Sparse Vector Technique. |
개인저자 | Chen, Yan. |
단체저자명 | Duke University. Computer Science. |
발행사항 | [S.l.]: Duke University., 2018. |
발행사항 | Ann Arbor: ProQuest Dissertations & Theses, 2018. |
형태사항 | 139 p. |
기본자료 저록 | Dissertation Abstracts International 79-10B(E). Dissertation Abstract International |
ISBN | 9780355892017 |
학위논문주기 | Thesis (Ph.D.)--Duke University, 2018. |
일반주기 |
Source: Dissertation Abstracts International, Volume: 79-10(E), Section: B.
Adviser: Ashwin Machanavajjhala. |
요약 | In today's fast-paced developing digital world, a wide range of services such as web services, social networks, and mobile devices collect a large amount of personal data from their users. Although sharing and mining large-scale personal data c |
요약 | Differential privacy has emerged as a de facto standard for analyzing sensitive data with strong provable privacy guarantees for individuals. There is a rich literature that has led to the development of differentially private algorithms for num |
요약 | First, we revisit the original Sparse Vector Technique and its variants, proving that many of its variants violate the definition of differential privacy. Furthermore, we design an attack algorithm demonstrating that an adversary can reconstruct |
요약 | Next, we utilize the original Sparse Vector Technique primitive to design new solutions for practical problems. We propose the first algorithms to publish regression diagnostics under differential privacy for evaluating regression models. Speci |
요약 | We then make use of Sparse Vector Technique as a key primitive to design a novel algorithm for differentially private stream processing, supporting queries on streaming data. This novel algorithm is data adaptive and can simultaneously support m |
일반주제명 | Computer science. |
언어 | 영어 |
바로가기 | ![]() |