Salsa Fresca: Angular Embeddings and Pre-Training for ML Attacks on Learning With Errors
This work improves attacks on post-quantum cryptography systems, which is significant for cryptanalysis and security evaluation, though it appears incremental as it builds on prior ML-based attacks.
The paper tackles the problem of slow and sample-inefficient machine learning attacks on Learning with Errors (LWE) cryptography by proposing three methods—better preprocessing, angular embeddings, and model pre-training—which speed up preprocessing by 25× and improve sample efficiency by 10×, enabling the first ML attack to recover sparse binary secrets in dimension n=1024.
Learning with Errors (LWE) is a hard math problem underlying recently standardized post-quantum cryptography (PQC) systems for key exchange and digital signatures. Prior work proposed new machine learning (ML)-based attacks on LWE problems with small, sparse secrets, but these attacks require millions of LWE samples to train on and take days to recover secrets. We propose three key methods -- better preprocessing, angular embeddings and model pre-training -- to improve these attacks, speeding up preprocessing by $25\times$ and improving model sample efficiency by $10\times$. We demonstrate for the first time that pre-training improves and reduces the cost of ML attacks on LWE. Our architecture improvements enable scaling to larger-dimension LWE problems: this work is the first instance of ML attacks recovering sparse binary secrets in dimension $n=1024$, the smallest dimension used in practice for homomorphic encryption applications of LWE where sparse binary secrets are proposed.