Paper 2020/1620

Neural Aided Statistical Attack for Cryptanalysis

Yi Chen and Hongbo Yu

Abstract

At CRYPTO 2019, Gohr proposed a neural aided attack on 11-round Speck32/64, which is the first work of neural aided cryptanalysis that is competitive to the state-of-the-art attacks against reduced versions of modern block ciphers. But such an attack can only work when there are plenty of neutral bits and relies purely on experiments for complexity evaluations. In this paper, we propose a neural aided statistical attack that almost can be as generic as the differential cryptanalysis. It has no special requirements about the attacked cipher and allows us to estimate the theoretical complexities and success rate. For reducing the key space to be searched, we propose a Bit Sensitivity Test to identify which ciphertext bit is informative. Then specific key bits can be recovered by building neural distinguishers on related ciphertext bits. Applications to round reduced Speck32/64, DES prove the correctness and superiorities of our neural aided statistical attack.