I am very grateful to receive the Netidee scholarship to support the completion of my PhD dissertation, “Advancing Privacy in Federated Learning.” This funding enables me to deepen my research into one of today’s most critical challenges in machine learning: ensuring strong data privacy without compromising model performance.

Federated Learning (FL) and Differential Privacy (DP) are two foundational technologies for privacy-preserving AI. FL enables models to be trained collaboratively without moving raw data, while DP provides formal guarantees that individual information cannot be reverse-engineered from model updates. However, stronger privacy protection typically leads to reduced model accuracy, especially in real-world settings where data across users is highly diverse and non-IID.

My research focuses on closing this gap. The goal of the project is to develop new methods that improve the privacy–utility trade-off in Federated Learning. By better integrating DP mechanisms into the FL process, the project aims to make it possible to train AI systems that are both private and effective, even under challenging data distributions.

A central contribution of this work will be the creation of open-source, publicly available tools and reproducible workflows. These resources will support developers, researchers, and organizations in adopting privacy-preserving techniques more easily and responsibly. Beyond academic impact, the project aims to empower the broader AI community to implement privacy by design in real applications — particularly in domains such as healthcare, where sensitive data is at the core of innovation.

With the support of Netidee, I will conduct extensive experimental evaluations, compare the proposed methods to state-of-the-art baselines, and publish the resulting insights as part of my dissertation. All scientific outputs, code, and documentation will be openly accessible to contribute to transparent and trustworthy AI innovation.

This work aligns strongly with Netidee’s mission of fostering an open, secure, and privacy-respecting digital ecosystem. I am thankful for the opportunity to advance research that supports responsible AI development in Austria and beyond.