Thesis supervisor: Zsolt Csaba Johanyák
Location of studies (in Hungarian): Óbuda University - Bánki Donát Faculty of Mechanical and Safety Engineering Abbreviation of location of studies: ÓEBGK
Description of the research topic:
Federated learning emerges as a promising solution to the challenges posed by traditional machine learning methods, particularly in handling vast volumes of sensitive data. By executing learning algorithms and models directly on client devices, such as smartphones and IoT devices, federated learning mitigates the risks associated with centralized data storage while preserving data privacy and protection. However, the decentralized nature of federated learning introduces new security concerns. This research endeavors to scrutinize the security risks inherent in federated learning
methodologies, with a focus on data privacy breaches and data integrity compromises. Through rigorous analysis and innovative methodologies, the aim is to devise robust mechanisms for the timely detection and prevention of malicious activities within federated learning ecosystems.
Research goals:
1. Analysis of security vulnerabilities in federated learning frameworks (e.g. model poisoning attacks, data leakage, inference attacks).
2. Development of new protection mechanisms and robust learning algorithms to increase the resilience of federated learning models against attacks.
3. Investigation of different techniques (e.g. differential privacy, homomorphic encryption, secure multi-agent computation) to reduce privacy risks in federated learning solutions.
4. Examination of the impact of federated learning security mechanisms on model performance (e.g. convergence, communication overheads).
5. Evaluate the proposed security solutions through empirical experiments and real case studies.
Required language skills: English B2 Number of students who can be accepted: 1
Deadline for application: 2024-08-31
2024. IV. 17. ODT ülés Az ODT következő ülésére 2024. június 14-én, pénteken 10.00 órakor kerül sor a Semmelweis Egyetem Szenátusi termében (Bp. Üllői út 26. I. emelet).
Login
