TRUMPET

One objective of the project focuses on developing a novel privacy metric and measurement tools particularly tailored to FL (Federated Learning) implementations. Project's efforts include the design of information-theoretic privacy metrics able to quantify risks such as data leakage and re-identification, while balancing privacy with model accuracy. These metrics will be validated into the FL TRUMPET platform, accompanied by PET algorithms. Through standardisation efforts, the project aims to explore measurable, transparent, and certifiable privacy practices to foster trust and enable cross-border data collaboration within FL frameworks.

Reason for applying to HSbooster.eu services

The project is applying for the Standardisation Booster service to ensure that its outcomes align with and contribute to international standards in privacy for Federated Learning (FL). Its goal is to establish an adequate privacy metric and certification methods to evaluate the privacy risks in FL, addressing current uncertainties in data protection practices and facilitating cross-border data collaboration. The Booster service can help by guiding it in aligning its metrics and tools with existing standards, identifying gaps where new standards are needed, and fostering collaboration with standardisation bodies. Additionally, the service can support it in promoting the adoption of its methodology, methods and tools within industry and regulatory frameworks, ensuring their practical impact and sustainability.