Paper 2020/571

Machine Learning Assisted Differential Distinguishers For Lightweight Ciphers

Anubhab Baksi and Jakub Breier and Xiaoyang Dong and Chen Yi

Abstract

At CRYPTO 2019, Gohr first introduces the deep learning based cryptanalysis on round-reduced SPECK. Using a deep residual network, Gohr trains several neural network based distinguishers on 8-round SPECK-32/64. The analysis follows an `all-in-one' differential cryptanalysis approach, which considers all the output differences effect under the same input difference. Usually, the all-in-one differential cryptanalysis is more effective than that only uses one single differential trail. However, when the cipher is non-Markov or its block size is large, it is usually very hard to fully compute. Inspired by Gohr's work, we try to simulate the all-in-one differentials for such non-Markov ciphers through deep learning. As proof of works, we trained several distinguishing attacks following machine learning simulated all-in-one differential approach. We present 8-round differntial distinguishers for Gimli-Hash and Gimli-Cipher, each with trivial complexity. Finally, we explore more on choosing an efficient machine learning model and show a three layer neural network can be used.