Awesome FL

Federated learning has gained significant attention as a privacy-preserving alternative to centralized training, particularly in healthcare, finance, and mobile applications where data cannot be shared due to privacy regulations, competitive concerns, or data residency requirements.

The collection is organized by research theme: communication efficiency (reducing the bandwidth cost of distributed training), aggregation algorithms (FedAvg, FedProx, SCAFFOLD, and their variants), personalization methods for heterogeneous data distributions, privacy-preserving techniques (differential privacy, secure aggregation, homomorphic encryption for gradients), and system challenges (client availability, stragglers, and heterogeneous hardware).

Links to major frameworksPySyft, Flower, FedML, TensorFlow Federatedare included alongside theoretical papers that underpin each research direction.

Researchers entering the federated learning field use this collection to orient themselves within the literature and identify the most important papers in each subarea.

ML engineers building federated learning systems for real-world deploymentmedical imaging across hospital networks, keyboard prediction on mobile devices, fraud detection across banking institutionsuse it to find implementations and frameworks relevant to their production constraints.

Academic groups publishing new federated learning methods use it as a venue to gain visibility for their work within the research community through community-contributed additions.