# A Blockchain-based Decentralized Federated Learning Framework with Committee Consensus

**Authors:** Yuzheng Li, Chuan Chen, Nan Liu, Huawei Huang, Zibin Zheng, Qiang Yan  
**arXiv:** [2004.00773](https://arxiv.org/abs/2004.00773)  
**Date:** 2020-04-02  
**Summary page:** https://www.summarizepaper.com/en/paper/2004.00773v1/

## Key points

- Federated learning is a popular approach for training machine learning models in various scenarios, including mobile computing.
- The security of federated learning is being questioned due to the potential for malicious clients or central servers to attack the global model or user privacy data.
- A new decentralized federated learning framework based on blockchain technology has been proposed to address these security concerns.
- The Blockchain-based Federated Learning framework with Committee consensus (BFLC) uses blockchain for global model storage and local model update exchange.
- BFLC includes an innovative committee consensus mechanism that reduces the amount of consensus computing required and minimizes the risk of malicious attacks.
- Scalability considerations such as theoretical security, storage optimization, and incentives are discussed in detail.
- Real-world experiments using datasets confirm the effectiveness of this new framework's approach to secure federated learning.

## Summary

Federated learning has become a popular approach for training machine learning models in various scenarios, including mobile computing. However, the security of federated learning is increasingly being questioned due to the potential for malicious clients or central servers to attack the global model or user privacy data. To address these security concerns, a new decentralized federated learning framework based on blockchain technology has been proposed. This Blockchain-based Federated Learning framework with Committee consensus (BFLC) uses blockchain for global model storage and local model update exchange. The BFLC also includes an innovative committee consensus mechanism that reduces the amount of consensus computing required and minimizes the risk of malicious attacks. The scalability of BFLC is discussed, including theoretical security, storage optimization, and incentives. Real-world experiments using datasets have been conducted to verify the effectiveness of this new framework. The authors propose a decentralized federated learning framework based on blockchain technology called BFLC that addresses security concerns related to malicious clients or central servers attacking global models or user privacy data. The BFLC uses blockchain for global model storage and local model update exchange while incorporating an innovative committee consensus mechanism that reduces consensus computing requirements and minimizes malicious attacks' risks. Scalability considerations such as theoretical security, storage optimization, and incentives are also discussed in detail. Finally, real-world experiments using datasets confirm the effectiveness of this new framework's approach to secure federated learning.