2018/04/11 by Yue Wang, Wang, Yue, Daniel Kifer +4 · 3 citations
Computer Science · Engineering · Mathematics · #Algorithm #Artificial intelligence #Computer science #Confidence interval #Cryptography and Data Security #Cryptography and Security (cs.CR) #Data collection #Data mining #Differential (mechanical device) #Differential privacy #Empirical risk minimization #Engineering #FOS: Computer and information sciences #Machine Learning (cs.LG) #Machine Learning (stat.ML) #Machine learning #Mathematics #Minification #Mobile Crowdsensing and Crowdsourcing #Noise (video) #Noisy data #Perturbation (astronomy) #Privacy-Preserving Technologies in Data #Sampling (signal processing) #Statistics #Synthetic data #Variety (cybernetics) #cs.CR #cs.LG #stat.ML
paper · pdf · doi:10.48550/arxiv.1804.03794
published in arXiv (Cornell University) (Cornell University)
arxiv created 2018/04/11 · openalex publication_date 2018/04/11 · arxiv updated 2018/04/12 · openalex created_date 2018/04/24 · openalex updated_date 2026/08/05
The process of data mining with differential privacy produces results that are affected by two types of noise: sampling noise due to data collection and privacy noise that is designed to prevent the reconstruction of sensitive information. In this paper, we consider the problem of designing confidence intervals for the parameters of a variety of differentially private machine learning models. The algorithms can provide confidence intervals that satisfy differential privacy (as well as the more recently proposed concentrated differential privacy) and can be used with existing differentially private mechanisms that train models using objective perturbation and output perturbation.