DC8 Project: Privacy-Preserving Federated Generative Models for Decentralised Data Synthesis

PhD Supervisor: prof. Francesco Piccialli (UNINA); Auxiliary supervisors: David Camacho (UPM), Dariusz Mrozek (SUT), Jia-Chun Lin (NTNU); R&D cooperation: CONFORM

Objectives: The primary objective of this research project is to design and validate generative models that can synthesise data in a federated learning environment that emphasises the preservation of privacy. As modern digital interactions become more pervasive, there is an increasing demand for models that can operate on decentralised data while ensuring users' privacy. This park of the project will attempt to address this by combining the potential of generative models with the decentralised nature of federated learning. The aim of the researcher will be to understand the challenges that arise when attempting to generate data from multiple decentralised sources, while also ensuring that the generated data does not compromise the privacy of any individual source. Another key objective will be to evaluate the quality and authenticity of the synthesised data and ensuring that it is comparable to what centralised generative models would produce.

Expected Results: At the culmination of this research, we anticipate a robust federated generative model that will be able to synthesise high-quality data across decentralised nodes without compromising individual data privacy. This model would be benchmarked against traditional centralised generative models in order to ensure that the quality of the synthesised data is maintained. Furthermore, the research is expected to produce a comprehensive understanding of the challenges and intricacies of data synthesis in a federated setting. This would include insights into any potential threats to privacy and methods to mitigate them. Additionally, the project is likely to yield protocols and/or best practices for developing and deploying federated generative models in various domains, from healthcare to finance, that will set the standard for future endeavours in this field.

Applied research: The applied research of this project focuses on developing a federated generative model prototype for data synthesis in privacy-sensitive sectors like healthcare and finance. It involves creating, testing, and refining the model using real-world datasets to ensure realistic data synthesis compliant with privacy regulations. The project aims to establish deployment strategies and best practices for integrating these models into various industry settings, addressing challenges like network latency and data heterogeneity, ensuring the model's practical viability and scalability in real-world applications.

Planned secondments: UPM(4 months); SUT(4 months); NTNU (4 months)                   

Enrolment in Doctoral degree: UNINA

Recruitment of Doctoral Candidate: https://euraxess.ec.europa.eu/jobs/290127