|
|
 |
| |
|
|
|
A Statistical Analysis of Commutativity Degrees in Finite Group Chains: Implications for Machine Learning–Based Cryptanalysis |
|
|
|
PP: 947-954 |
|
|
doi:10.18576/jsap/140630
|
|
|
|
Author(s) |
|
|
|
Sulieman Shelash,
Hamza Farhan Ahmad,
A. Vasudevan,
Hanan Jadallah,
Mohammad Faleh Hunitie,
Yogeesh N,
|
|
|
|
Abstract |
|
|
| In this work, we introduced the graph-theoretic vulnerability modelling framework of post-quantum cryptosystems
with the deep-learning-based performance evaluation is introduced. The suggested solution combines structural graph modelling of cryptographic elements with regression, and classification-based predictive analytics to determine quantitatively how resilient a system will be in terms of its attack surface in the quantum era. The regression metrics MSE, RMSE, MAE and R 2 metrics quantify the prediction fidelity of structural risk estimation and classification metrics accuracy, precision, recall, F1-score and ROC-AUC metrics finally allow accurate detection of vulnerable states of cryptographic settings in a variety of different cryptographic configurations. This methodology shows that the deep learning models are applicable to predicting the probability of vulnerabilities with graph-based cryptographic features, and it provides a scalable analysis pipeline with the use of emerging post-quantum technologies. Findings validate the hypothesis that the graph-theoretic and machine-learning framework is highly beneficial in enhancing the robustness analysis of lattice-, hash-, and code-based
encryption, which in turn can be used to assist more secure and resilient cryptographic constructions. |
|
|
|
|
|
|
|
