The recent advancements in federated learning (FL) have significantly focused on enhancing privacy, security, and efficiency. A notable trend is the development of novel frameworks that address privacy concerns without compromising model performance. Techniques such as differential privacy, secure multi-party computation, and innovative communication strategies are being integrated to create lightweight and robust solutions. Additionally, there is a growing emphasis on mitigating specific privacy threats like gradient inversion attacks and membership inference attacks through innovative architectural designs and advanced cryptographic methods. These developments not only aim to protect sensitive data but also to ensure faster convergence and resilience against stragglers and malicious actors. Furthermore, the field is witnessing a shift towards decentralized training paradigms that promise greater computational resources and democratized access, albeit with new challenges such as governance and the 'No-Off Problem'. Overall, the research direction in FL is moving towards more secure, private, and efficient collaborative learning environments.
Enhancing Privacy and Security in Federated Learning
Sources
Communication Compression for Distributed Learning without Control Variates
Privacy Drift: Evolving Privacy Concerns in Incremental Learning
Lightweight Federated Learning with Differential Privacy and Straggler Resilience
Membership Inference Attacks and Defenses in Federated Learning: A Survey
A New Federated Learning Framework Against Gradient Inversion Attacks
Tazza: Shuffling Neural Network Parameters for Secure and Private Federated Learning
Adaptive Epsilon Adversarial Training for Robust Gravitational Wave Parameter Estimation Using Normalizing Flows
Protocol Learning, Decentralized Frontier Risk and the No-Off Problem
How Does the Smoothness Approximation Method Facilitate Generalization for Federated Adversarial Learning?
Multi-client Functional Encryption for Set Intersection with Non-monotonic Access Structures in Federated Learning