Title:	Analysis of Post quantum cryptography using neural estimators of mutual information
DNr:	LiU-compute-2025-45
Project Type:	LiU Compute
Principal Investigator:	Onur Günlü <onur.gunlu@liu.se>
Affiliation:	Linköpings universitet
Duration:	2025-11-07 – 2026-01-01
Classification:	10211
Keywords:

Abstract

This study utilizes the training of neural networks, where one network is trained for each cryptographic algorithm in the study. One network training takes on average 1.45 *1000 cpu hours, and to finish this study we have about 10-15 networks left to train. Given that we have about two months more to finish this study we would like to request that the allocated be increased to 15* 1000 cpu hours to have time to finish the study in time and some margin in something unforeseen happens. The purpose of this work is to analyze the security of candidate algorithms for Post Quantum cryptography (PQC), and hybrid PQC methods. This will be done by estimating the mutual information between ciphertexts and plaintext. This estimation will be done using neural estimators, primarily Mutual information neural estimator (MINE).