Pruning-Aware CryptoKAN Models for Privacy-Preserving Intrusion Detection

This project explores how advanced artificial intelligence models can analyze network traffic while keeping sensitive data fully encrypted. Building on recent research using CryptoKAN neural networks, the project studies whether model compression techniques can make encrypted intrusion detection faster and more practical without reducing detection accuracy. The intern will first adapt and evaluate encrypted AI models for detecting cyberattacks, then apply pruning methods to reduce their computational cost. The results will help academic researchers better understand the trade-offs between privacy, security, and efficiency, while providing practical insights for institutions seeking secure data analysis solutions that do not expose confidential information.

Faculty Supervisor:

Chamseddine Talhi

Student:

Partner:

École Nationale Polytechnique

Discipline:

Computer science

Sector:

Artificial Intelligence; Cyber Security; Information and Communications Technology (ICT)

University:

École de technologie supérieure

Program: