The field of federated learning is rapidly advancing, with a strong focus on improving security and privacy. Recent developments have highlighted the importance of protecting against gradient inversion attacks, malicious clients, and backdoor attacks. Researchers are proposing innovative defense mechanisms, such as shadow modeling, dimensionality reduction, and reputation systems, to mitigate these threats. Additionally, there is a growing interest in developing frameworks that can detect and prevent malicious behavior, such as anomaly detection and intrusion detection systems. The use of decentralized finance platforms and automated market makers is also being explored to create more flexible and scalable reward distribution systems. Notable papers in this area include SecureFed, which presents a two-phase framework for detecting malicious clients, and SPA, which proposes a novel backdoor attack framework that leverages feature-space alignment. These advancements demonstrate the field's commitment to addressing the unique security challenges introduced by federated learning and ensuring the privacy and integrity of sensitive data.
Advances in Federated Learning Security and Privacy
Sources
Shadow defense against gradient inversion attack in federated learning
SecureFed: A Two-Phase Framework for Detecting Malicious Clients in Federated Learning
TriCon-SF: A Triple-Shuffle and Contribution-Aware Serial Federated Learning Framework for Heterogeneous Healthcare Data
Behavioral Anomaly Detection in Distributed Systems via Federated Contrastive Learning
Network Structures as an Attack Surface: Topology-Based Privacy Leakage in Federated Learning
RepuNet: A Reputation System for Mitigating Malicious Clients in DFL
A Hybrid Intrusion Detection System with a New Approach to Protect the Cybersecurity of Cloud Computing
Can One Safety Loop Guard Them All? Agentic Guard Rails for Federated Computing
WallStreetFeds: Client-Specific Tokens as Investment Vehicles in Federated Learning
Hear No Evil: Detecting Gradient Leakage by Malicious Servers in Federated Learning
SPA: Towards More Stealth and Persistent Backdoor Attacks in Federated Learning