Publication from Master Thesis Supervision (Espen Sele)

Published: December 22, 2025

A work related to the Master thesis of Espen Sele, co-supervised with Prof. Ferhat Ozgur Catak, was published.

Published Title

Secure federated learning via neural cryptography with homomorphic operations

Authors: Espen Sele, Ferhat Ozgur Catak, Jungwon Seo, Murat Kuzlu

Brief Summary

The paper studies neural cryptography with homomorphic operations as a secure aggregation alternative for federated learning. It introduces a neural homomorphic operation system and reports competitive results on MNIST (including 88.10% accuracy with quantized weights), showing its potential as a lightweight option compared with traditional homomorphic encryption schemes.

• Publication Page