A Test for the Two-Sample Problem using Mutual Information to Fix Information Leak in e-Passports Seminar

For two independently drawn samples from continuous distributions, a statistical test based on the mutual information statistic is proposed for the null hypothesis that both samples originate from the same population. We demonstrated through simulation that this test is more powerful than the commonly used nonparametric tests in many situations.

As an application we discuss the information leak of e-passports. Using the mutual information based tests, we show that it is possible to trace a passport by comparing the time taken to respond to a recorded message. We show that the mutual information based test detects a leak in a situation where the other non-parametric tests fail.

This is a joint work with Tom Chothia, University of Birmingham.