The recent advancements in privacy-preserving technologies have significantly influenced various fields, particularly in healthcare and human-robot interaction. Federated learning (FL) has emerged as a cornerstone, enabling collaborative model training across decentralized data sources while maintaining strict privacy. This approach has been particularly impactful in medical imaging and brain-computer interfaces, where the need for large datasets and stringent privacy requirements are paramount. Innovations in FL, such as the integration of differential privacy and secure aggregation, have demonstrated near-equivalent performance to non-private models, addressing critical privacy concerns in healthcare data management. Additionally, advancements in homomorphic encryption and synthetic data generation have further fortified privacy-preserving practices, offering new avenues for secure data sharing and analysis. Notably, the application of FL in human-robot interaction has shown promising results in complex command classification and occlusion prevention, enhancing the practicality of multi-robot systems. These developments collectively underscore a shift towards more secure, efficient, and collaborative data-driven solutions across diverse domains.
Privacy-Preserving Innovations in Federated Learning and Data Security
Sources
Sharing the Path: A Threshold Scheme from Isogenies and Error Correcting Codes
Towards Privacy-Preserving Medical Imaging: Federated Learning with Differential Privacy and Secure Aggregation Using a Modified ResNet Architecture
Federated Motor Imagery Classification for Privacy-Preserving Brain-Computer Interfaces
HumekaFL: Automated Detection of Neonatal Asphyxia Using Federated Learning
Linearly Homomorphic Signature with Tight Security on Lattice
Privacy-Preserving Federated Learning via Homomorphic Adversarial Networks
Proximal Control of UAVs with Federated Learning for Human-Robot Collaborative Domains
End to End Collaborative Synthetic Data Generation
Privacy-Preserving in Medical Image Analysis: A Review of Methods and Applications
Multi-Layer Privacy-Preserving Record Linkage with Clerical Review based on gradual information disclosure