Computer Science: Fully Funded PhD Scholarship: Graduate Teaching Assistantship for PhD studies in Computer Science (RS699)

Funding provider(s): Swansea University  

Subject area(s): Cyber Security 

Project start date(s): 

• 1 January 2025 (Enrolment open from mid-December) 

• 1 April 2025 

Supervisors: Dr. Yang Liu, Dr Lu Zhang, Professor Markus Roggenbach 

Aligned programme of study: 

PhD in Computer Science, Extended Period of Study [UK students] 
PhD Extended Period of Study – Swansea University [International students] 

Mode of study: Full-time 

Project description: 

The Department of Computer Science is offering Graduate Teaching Assistantships for January 2025 entry to its PhD studies in Computer Science. This position is ideal for candidates with a keen interest in pursuing doctoral research in cyber security, particularly focusing on privacy, security, and robustness in Federated Learning (FL). Successful applicants will also have the opportunity to gain valuable teaching training and experience while developing their professional skills. The GTA will complete the Associate Fellowship of Advance HE (AFHEA) and thereby develop their teaching related skills and enhance their professional competencies. The equivalent of 3 months per year will be devoted to teaching training/practice, the remaining time is for research studies on a four year programme. 

Research Topic: Privacy and Robustness in Federated Learning 

As the deployment of Federated Learning (FL) expands across various domains, so do the challenges related to privacy, security, and robustness. Its distributed nature introduces unique vulnerabilities that traditional centralized learning approaches do not face. Attacks on FL can result in significant privacy breaches, compromising sensitive data across multiple participants.  

This PhD project focuses on the exploration and analysis of threats, attacks, and defences within FL environments. The aim is to build a comprehensive understanding of the vulnerabilities in FL systems, design robust countermeasures, and establish guidelines for secure and privacy-preserving FL implementations. The project will not only focus on theoretical advancements but also emphasize practical applications in real-world scenarios. The successful candidates will have the opportunity to collaborate with a multidisciplinary team, including experts from both academia and industry, to ensure the research outcomes are both innovative and impactful. The ideal candidate should be passionate about tackling complex challenges at the intersection of AI, privacy, and security , and eager to contribute to both academic research and teaching. 

Key Objectives: 

1. Investigate Federated Learning Threats: Conduct a deep dive into potential attacks on FL systems, including poisoning attacks, backdoor attacks, and inference attacks. Understand how these attacks exploit the distributed nature of FL to breach privacy and compromise model integrity.
2. Design Defensive Mechanisms: Develop advanced defence strategies against the identified attacks. This includes exploring techniques like differential privacy, secure aggregation, and adversarial training to enhance the robustness and privacy of FL models.
3. Real-world Application and Validation: Implement and test the developed models on real-world datasets from domains such as healthcare, finance, or IoT. Validate the effectiveness of the proposed defences in preserving privacy and robustness under adversarial conditions.

For more details please see here: https://www.swansea.ac.uk/postgraduate/scholarships/research/gta-phd-computer-science-rs699.php