Privacy-Preserving Federated Learning for Phishing Detection

July 18th, 2022

Thesis Type Bachelor
Student Farnaz Khorsand
Status Running
Proposal on 13/09/2022 10:45 am
Proposal room Seminar room I5 6202
Supervisor(s) Stefan Decker
Advisor(s) Mehdi Akbari G.

With the irreversible advancements of technology, strategies are used to commit cybercrimes to improve equally. Thus, in the context of cybercrime mitigation, novel technologies often draw the attention of research and cyber security teams. Phishing is one of the oldest yet most present and known types of cybercrime committed with the goal of information theft. The increasing interconnectivity of data across cyberspace intensifies concerns regarding data privacy and users’ data privacy protection. Machine learning and its growing subfields are novel but empirically successful approaches for combatting classification and detection issues. This work focuses on a particular form of privacy-preserving machine learning, precisely Secure Multi-Party Computation (SMPC)-based federated learning aiming to evaluate the feasibility of this privacy-preserving machine learning approach for real-world use cases with regard to existing frameworks and tools.


Evaluation of SMPC-based federated-learning libraries for phishing classification.