Side-channel attack on the human brain : facial recognition and preferences

The term "side-channel attack" in computer system describes the fact that an attacker tries to get information about the functioning of the whole system by analyzing features like timing information, power consumption or electromagnetic leaks. Applied to the human brain, this means that a researcher tries to obtain hidden information on a subject by analyzing his EEG signals. Several studies already showed that it was possible to use some events presents on these EEG segments to determine if a subject is lying, find his PIN code, his bank,... In this thesis, we try to demonstrate the difficulty of retrieving such information with non-optimal experimental conditions. For this purpose, we designed two experiments based respectievely on face recognition and preference of users.