UTFacultiesEEMCSDisciplines & departmentsSCSEducationAssignmentsFinished AssignmentsFinished Master AssignmentsWeighted, Weighted and Art Found Wanting: A Complexity-minimisation Approach for Neuroevolution-based Side-channel Analysis

Weighted, Weighted and Art Found Wanting: A Complexity-minimisation Approach for Neuroevolution-based Side-channel Analysis

MASTER Assignment

Weighted, Weighted and Art Found Wanting: A Complexity- minimisation Approach for Neuroevolution-based Side-channel Analysis

Type : Master M-CS

Period: August, 2024 - January, 2025

Student : Velde, van der P. (Peter, Student M-CS)

Date Final project: January 27, 2025

Thesis

Supervisors:

Abstract:

Currently the state of the art in Side-Channel Analysis in the sphere of cryptography is to analyze them using Deep Neural Networks (DNN). A common problem in this field is to minimize both the number of traces required to reach a good classification performance and the number of trainable parameters of the DNN. Recently a neuroevolution approach was researched as a possible solution to this problem called NASCTY. With this research project we hope to discover a number of possible improvements to the current system hoping to overcome some of its deficits and problems. This includes looking at a custom fitness function to reduce complexity and the use of different anti-premature convergence strategies.