Research Fellow - Federate Learning Security

Federated learning (FL) is a privacy-preserving distributed learning paradigm that allows different clients to create a shared AI models without having to share their data. Despite these advantages, conventional FL frameworks are not exempt of security weaknesses. The role involves performing research to create on decentralised federated learning that 1) improved robustness against poisoning and adversaries whilst 2) while maintaining their privacy preserving properties.

Specifically, the role will involve investigating:

• Decentralised FL convergence and vulnerabilities
• Novel decentralised robust aggregation technique considering local consensus by clustering the nodes basedon similar data distribution
• Novel decentralised robust aggregation technique considering local consensus by clustering the nodes basedon local trust and consensus

About the person:

To be considered for interview, applicants must demonstrate that they meet the essential criteria for the post, which includes:

• Have or about to obtain a PhD in computer science, engineering, mathematics or physical sciences area
• Recent high quality research experience in machine learning/AI, or both, as evidenced by a strong track record of publications in leading journals and conferences in relevant areas.
• Software programming skills

To be successful at shortlisting stage, please ensure you clearly evidence in your application how you meet the essential and, where applicable, desirable criteria listed in the Candidate Information document on our website.

Advert information