Differential Privacy

A privacy protection framework requiring that a mechanism produce similar output distributions for adjacent datasets, hiding the influence of a single record through added noise. Its core metrics include global sensitivity and local sensitivity.