Jacobian-guided reshaping converts isotropic LDP noise into an anisotropic distribution focused on task-relevant subspaces, yielding roughly 20% utility gains on CIFAR-10-C for PrivUnit2 and PrivUnitG at ε=7.5 while keeping per-dimension privacy budgets uniform.
Collecting and Analyzing Data from Smart Device Users with Local Differential Privacy
3 Pith papers cite this work. Polarity classification is still indexing.
3
Pith papers citing it
abstract
Organizations with a large user base, such as Samsung and Google, can potentially benefit from collecting and mining users' data. However, doing so raises privacy concerns, and risks accidental privacy breaches with serious consequences. Local differential privacy (LDP) techniques address this problem by only collecting randomized answers from each user, with guarantees of plausible deniability; meanwhile, the aggregator can still build accurate models and predictors by analyzing large amounts of such randomized data. So far, existing LDP solutions either have severely restricted functionality, or focus mainly on theoretical aspects such as asymptotical bounds rather than practical usability and performance. Motivated by this, we propose Harmony, a practical, accurate and efficient system for collecting and analyzing data from smart device users, while satisfying LDP. Harmony applies to multi-dimensional data containing both numerical and categorical attributes, and supports both basic statistics (e.g., mean and frequency estimates), and complex machine learning tasks (e.g., linear regression, logistic regression and SVM classification). Experiments using real data confirm Harmony's effectiveness.
representative citing papers
Develops LDP perturbation mechanisms and conditional estimation methods for key-value data to enable privacy-preserving analysis of conditional probabilities and marginals.
Tutorial summarizing existing Local Differential Privacy algorithms for heavy hitter identification, spatial data collection, and open problems.
citing papers explorer
-
Jacobian-Guided Anisotropic Noise Reshaping for Enhancing Representation Utility under Local Differential Privacy
Jacobian-guided reshaping converts isotropic LDP noise into an anisotropic distribution focused on task-relevant subspaces, yielding roughly 20% utility gains on CIFAR-10-C for PrivUnit2 and PrivUnitG at ε=7.5 while keeping per-dimension privacy budgets uniform.
-
Conditional Analysis for Key-Value Data with Local Differential Privacy
Develops LDP perturbation mechanisms and conditional estimation methods for key-value data to enable privacy-preserving analysis of conditional probabilities and marginals.
-
Local Differential Privacy: a tutorial
Tutorial summarizing existing Local Differential Privacy algorithms for heavy hitter identification, spatial data collection, and open problems.