(updated on 20 November 2018)
Exploration of recurrent neural network in template attacks
Person of contact: Aymeric Genet ([email protected])
Explore the recurrent neural network as a profiling method to perform key recovery attacks on traditional cryptosystems (AES, …) using side-channel information.
Computing endomorphism rings of supersingular elliptic curves
Person of contact: Novak Kaluderovic ([email protected])
One of the proposed post-quantum protocols for public key cryptography is based on computing isogenies of supersingular elliptic curves over finite fields. Currently the best known attacks on this protocol are exponential both in the classical and quantum case. There is a correspondence between endomorphisms of supersingular elliptic curves and quaternion algebras whose efficient computation would lead to an attack of the scheme. The student’s goal is to further study this correspondence and implement it in a computer language of own choice.