Differential Privacy Tutorials

$$ \Pr[{\mathcal {A}}(D_{1})\in S] \leq e^{\varepsilon} \Pr[{\mathcal {A}}(D_{2})\in S] + \delta $$

I was fortunate to be named OpenMined’s contributor of the month for writing the following differential privacy tutorials:

(1) Choosing Epsilon

(2) Differential Identifiability

(3) Local Sensitivity from Scratch

(4) Global Sensitivity from Scratch