| There are several methodologies available for analyzing the security of cryptographic algorithms, each relying on the ability to represent the cipher using mathematical or logical expressions. However, this representation process is time-consuming. Therefore, especially in the preliminary stages of cipher analysis, it proves advantageous to employ methodologies that enable a faster, albeit approximated, security evaluation. In this context, this paper introduces the development of a comprehensive framework designed to facilitate such analyses through the application of machine learning (ML) techniques. The core motivation behind this approach lies in leveraging the inherent approximation capabilities of ML approaches, expected to significantly mitigate the computational cost associated with cipher analysis. By exploiting the power of ML, the proposed framework seeks to provide an initial, yet insightful assessment of the security landscape of block ciphers, particularly during the preliminary stages of their evaluation. |
*** Title, author list and abstract as seen in the Camera-Ready version of the paper that was provided to Conference Committee. Small changes that may have occurred during processing by Springer may not appear in this window.