Séminaire Calcul Formel

Attacking LWE-based cryptosystems with sparse binary secrets using Transformers

by Dang-Hung DUONG

XR 203 (XLIM)

XR 203



Learning With Errors (LWE), one of the most fundamental foundations in lattice-based cryptography, plays a crucial role in Post-Quantum Cryptography (PQC) as several recent standardizations by the NIST such as CRYSTALS-Kyber and CRYSTALS-Dilithium are built upon it. However, recent publications by E. Wenger, C. Li et al. in 2023: SALSA and SALSA PICANTE, have proven that Machine Learning can be used to solve LWE problems with sparse binary secrets, therefore potentially threatening the security of LWE-based cryptosystems. This presentation mainly focuses on studying and analyzing the methods proposed.