As part of the EU’s Marie Skłodowska-Curie Actions (MSCA), Mr. Thanh Linh Nguyen from Trinity College Dublin (TCD) has successfully completed a three-month research secondment at Delft University of Technology (TU Delft), Netherlands. The exchange, facilitated by the ENSURE-6G Staff Exchange Program, focused on developing a trustworthy federated unlearning (FUL) framework that can verifiably erase user data and its associated influence from the trained AI model.
During the secondment, Linh worked under the supervision of Prof. Aaron Ding at TU Delft, collaborating closely with Prof. Marcela (TU Delft) and Prof. An from Vrije Universiteit Brussel, who was also on an ENSURE-6G exchange. Their research discussions focused on verifiable FUL—specifically, how to design a trustworthy FUL framework in which clients can verify that their data’s influence has been genuinely removed from both the training pipeline and the trained model. This collaboration has led to a draft magazine article, set for submission in October 2025. The article will propose a structured and comprehensive framework for verifiable FUL, which is expected to be among the first of its kind.
The research has broader implications for data privacy and the digital economy. It explores how to fairly compensate data contributors, apply cryptographic methods for trust, and ultimately allow users to monetize their data in a privacy-preserving way. This is especially vital for data-sensitive applications in regulated fields such as healthcare.
