Federated Learning Framework with Differential Privacy over Homomorphic Vector Encryption for Data-Sensitive Applications
Abstract
Abstract Federated Learning (FL) has become a popular paradigm in recent years, attributed to its concept of insight sharing for ensuring data privacy. It has been found to be of extensive utility in applications that regularly deal with confiden-tial data and has been widely used in the fields of healthcare and the Internet of Medical Things (IoMT). However, an equal amount of research is conducted to reverse-engineer the used datasets from the shared insights, and a native FL implementation alone is not suitable for these sensitive applications. We pro-pose a lightweight secure FL framework incorporating both Differential Privacy and Homomorphic Encryption at the vector level to minimize the efficiency of reverse-engineering algorithms over the transmitted insights in IoMT and other resource-constrained networks. Through experiments on multiple image classi-fication datasets and comparison with secure federated learning baselines, the proposed framework demonstrates the feasibility of combining differential pri-vacy with encrypted aggregation while quantifying the resulting predictive and cryptographic overhead.
Read the paper
Related studies
- Active Learning during Federated Learning for Object Detection
- Fairness-aware federated learning
- Decentralized Federated Learning
- Building a Personally Identifiable Information Recognizer in a Privacy Preserved Manner Using Automated Annotation and Federated Learning
- Preservation of Patient Level Privacy: Federated Classification and Calibration Models
- Efficient explainable federated learning
Go deeper in Evidence Hub
Evidence Hub adds a study data sheet (design, sample size, evidence level, key numbers), an AI summary and a chat grounded in the paper's text.