ConceptGlobal Sensitivity
Global Sensitivity
A concept in differential privacy that measures the maximum difference in a function's output over any adjacent datasets, that is, the worst-case influence. It is used to calibrate the noise scale of mechanisms such as the Laplace mechanism.
Timeline (last 90 days)
Oct 30
The global sensitivity of differential privacy measures the maximum difference in a function's output over any neighboring datasets, whereas local sensitivity is computed for a specific dataset and is usually far smaller than global sensitivity.
Unverified50%