학술행사
- 발표자 최호식 교수(서울시립대학교)
- 개최일시 2023-04-12 14:00-16:00
- 장소 국가수리과학연구소 산업수학혁신센터(광교)
-
일시: 2023년 4월 12일(수), 14:00-16:00
-
장소: 판교 테크노밸리 산업수학혁신센터 세미나실
- 경기 성남시 수정구 대왕판교로 815, 기업지원허브 231호 국가수리과학연구소
- 무료주차는 2시간 지원됩니다.
-
발표자: 최호식 교수(서울시립대학교)
-
주요내용: Quantile Estimation for Encrypted Data
Third-party analyzer which uses personal information may expose sensitive information. As data-based researches increase, such privacy issues are being raised seriously. Homomorphic encryption(HE) has been proposed as a way to avoid it. However, a challenge in ciphertext used in HE is a calculation time of elementary operations, which has significantly much higher complexity than that on plaintexts, resulting in that subsequent data analysis limiting various statistical analytics compared to plaintext data. In this paper, we consider an estimation method to get quantiles for encrypted data, where quantiles are core statistics to figure out data distribution in statistical data analysis. We propose a HE-friendly algorithm for large homomorphic encrypted data using an approximate quantile loss function. Numerical studies show that our proposed method improves calculation time over an estimation based on a sorting way for randomly generated and real data's homomorphically encrypted data. Furthermore, we suggest a modified boxplot for homomorphically encrypted data, where a boxplot is a primary statistical analysis method necessary for data analysis.
유튜브 실시간 스트리밍 : 현장 참석이 어려운 분들을 위해 온라인으로 실시간 방송할 예정입니다. 주소는 당일 신청 페이지에 업데이트 하겠습니다.
-
일시: 2023년 4월 12일(수), 14:00-16:00
-
장소: 판교 테크노밸리 산업수학혁신센터 세미나실
- 경기 성남시 수정구 대왕판교로 815, 기업지원허브 231호 국가수리과학연구소
- 무료주차는 2시간 지원됩니다.
-
발표자: 최호식 교수(서울시립대학교)
-
주요내용: Quantile Estimation for Encrypted Data
Third-party analyzer which uses personal information may expose sensitive information. As data-based researches increase, such privacy issues are being raised seriously. Homomorphic encryption(HE) has been proposed as a way to avoid it. However, a challenge in ciphertext used in HE is a calculation time of elementary operations, which has significantly much higher complexity than that on plaintexts, resulting in that subsequent data analysis limiting various statistical analytics compared to plaintext data. In this paper, we consider an estimation method to get quantiles for encrypted data, where quantiles are core statistics to figure out data distribution in statistical data analysis. We propose a HE-friendly algorithm for large homomorphic encrypted data using an approximate quantile loss function. Numerical studies show that our proposed method improves calculation time over an estimation based on a sorting way for randomly generated and real data's homomorphically encrypted data. Furthermore, we suggest a modified boxplot for homomorphically encrypted data, where a boxplot is a primary statistical analysis method necessary for data analysis.
유튜브 실시간 스트리밍 : 현장 참석이 어려운 분들을 위해 온라인으로 실시간 방송할 예정입니다. 주소는 당일 신청 페이지에 업데이트 하겠습니다.
