Dernière mise à jour :
A system for publicly sharing information about a dataset by describing patterns of groups within the dataset while withholding information about individuals.
Formalizes privacy guarantees by adding calibrated noise to query results (e.g., counts, means) so that the presence or absence of any single individual in the dataset cannot be inferred. Differential privacy parameters (ε, δ) quantify privacy loss, enabling organizations to balance data utility and individual protection, and must be managed centrally to track cumulative privacy budget.
Exemple concret :
A national statistics office releases census aggregates with differential-privacy noise injection. Researchers querying demographic counts receive slightly perturbed results, preserving overall trends while preventing re-identification of individual respondents - even when combined with other datasets.
