Gruppo di Logica e Geometria della Cognizione

Software systems may leak information about confidential data.
Information theory helps in mathematically analyze and automatically quantify such leaks.

However, there is an (uncountable) infinity of leakage measures.
Hence a question arises: can we design countermeasures to information leakage
that are universally optimal, i.e. that guarantee minimal leakage no matter
which leakage measure is chosen?

We show contexts when this can be achieved, a counterexample to a general
solution and discuss potential applications of these techniques to security and privacy problems.

(This is a joint work with Arman Khouzani)