The recent advancements in federated learning (FL) and related technologies have significantly enhanced privacy, security, and efficiency in decentralized data processing. A notable trend is the integration of hardware-based security mechanisms, such as exclaves and TrustZone, to ensure the integrity and transparency of FL processes. These innovations address critical vulnerabilities, including backdoor attacks and gradient leakage, by implementing novel defense strategies and model transformations. Additionally, the use of local differential privacy (LDP) has been extended to federated heavy hitter analytics, offering a robust solution for privacy-preserving data collection and analysis. In the realm of real-time applications, fall detection systems leveraging smartphone accelerometers and Wi-Fi CSI have shown exceptional accuracy and energy efficiency, marking a promising direction for eldercare technologies. Overall, the field is progressing towards more secure, efficient, and privacy-conscious solutions, with a strong emphasis on practical implementation and measurable performance improvements.
Enhancing Privacy, Security, and Efficiency in Federated Learning and Real-Time Applications
Sources
Real-Time Fall Detection Using Smartphone Accelerometers and WiFi Channel State Information
ExclaveFL: Providing Transparency to Federated Learning using Exclaves
Client-Side Patching against Backdoor Attacks in Federated Learning
Capacity of Hierarchical Secure Coded Gradient Aggregation with Straggling Communication Links
Just a Simple Transformation is Enough for Data Protection in Vertical Federated Learning
Efficiently Achieving Secure Model Training and Secure Aggregation to Ensure Bidirectional Privacy-Preservation in Federated Learning
Building Gradient Bridges: Label Leakage from Restricted Gradient Sharing in Federated Learning
Hybrid CNN-LSTM based Indoor Pedestrian Localization with CSI Fingerprint Maps
T-Edge: Trusted Heterogeneous Edge Computing
Federated Heavy Hitter Analytics with Local Differential Privacy