A Blockchain-Based Trustworthy Federated Learning Privacy Protection Scheme
Abstract
With the emergence of distributed machine learning, federated learning has become a prevalent paradigm for multi-party collaborative training scenarios. To address security vulnerabilities such as imperfect encryption and incapability to with-stand attacks within existing federated learning architectures, a blockchain-based trustworthy federated learning architecture is proposed. The architecture utilizes a threshold variant of the Paillier encryption scheme to protect the local and global model updates. A flexible Multi-Krum selection scheme is also proposed for balancing the utility of the overall architecture while ensuring safe federated learning training processes. Based on theoretical analysis and experimental results, the proposed architecture demonstrates a high level of security and the feasibility of the scheme is supported.
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.