Federated Learning has garnered much attention recently with its promises to extract insights from sensitive client data without compromising privacy, thus allowing for privacy-preserving collaborative learning on an ever-growing number of edge devices. However, despite the recent explosion in the research on federated learning, there is a disconnect between its theoretical underpinnings and its practical application. On the one hand, despite more than a decade of work, we still need an intimate understanding of what data properties lead to efficient federated optimization. On the other hand, recent work on designing privacy attacks has shown that merely keeping the data on a client's device doesn’t guarantee meaningful privacy, thereby putting the central promise behind the discipline in question. The primary goal of this workshop is to underline the limitations of these two aspects of optimization and privacy in federated learning. 

This workshop aims to underline the limitations of optimization and privacy aspects in federated learning. Additionally, we aim to discuss emerging works proposing novel privacy notions and guarantees specialized for federated settings. This is a crucial area of research to bolster confidence in the field and provide a much-needed direction to the community for future research. Finally, emerging data protection laws make it easier for client devices to leave the collaboration if and when they want to. Still, there is very little work in designing incentives for client participation in federated learning. We also intend to cover mechanism design and fairness topics in federated learning. 

We aim to showcase interesting works that suggest novel techniques to overcome real-world challenges. The workshop will feature several computer science, optimization, and statistics speakers to highlight recent progress, identify key challenges, and discuss future directions in Federated Learning. Panel discussions and social events will augment contributed talks.

Venue: TTIC Chicago, USA (Sept 8-9, 2023)