AutoDFL: A Scalable and Automated Reputation-Aware Decentralized Federated Learning

2025-01-08

AI Summary

Blockchain-based federated learning (BFL) faces critical challenges including on-chain congestion, low throughput, incentive imbalance, and coarse-grained reputation management. To address these, this paper proposes a Layer-2, reputation-aware BFL framework built upon zk-Rollups. Our method introduces an automated, dynamic, and verifiable reputation model that integrates decentralized identifiers (DIDs), off-chain computation verification, and dynamically weighted model aggregation. Leveraging zk-Rollups, we batch-compress reputation updates and global model aggregations onto-chain, drastically reducing on-chain footprint. Experimental results demonstrate an average throughput exceeding 3,000 transactions per second (TPS) and a 20× reduction in gas consumption. The framework maintains privacy, security, and transparency while significantly enhancing scalability, economic efficiency, and incentive fairness in BFL systems.

Technology Category

• Machine Learning: Distributed Machine Learning & Federated Learning
• Multiagent Systems: Mechanism Design
• Natural Language Processing: Safety and Robustness

Application Category

• Security and Privacy: Blockchains and distributed ledgers
• User Modeling, Personalization and Recommendation: Federated recommendation systems and personalization
• Responsible Web: Data and user privacy-enhancing technologies for the Web

Abstract

Blockchained federated learning (BFL) combines the concepts of federated learning and blockchain technology to enhance privacy, security, and transparency in collaborative machine learning models. However, implementing BFL frameworks poses challenges in terms of scalability and cost-effectiveness. Reputation-aware BFL poses even more challenges, as blockchain validators are tasked with processing federated learning transactions along with the transactions that evaluate FL tasks and aggregate reputations. This leads to faster blockchain congestion and performance degradation. To improve BFL efficiency while increasing scalability and reducing on-chain reputation management costs, this paper proposes AutoDFL, a scalable and automated reputation-aware decentralized federated learning framework. AutoDFL leverages zk-Rollups as a Layer-2 scaling solution to boost the performance while maintaining the same level of security as the underlying Layer-1 blockchain. Moreover, AutoDFL introduces an automated and fair reputation model designed to incentivize federated learning actors. We develop a proof of concept for our framework for an accurate evaluation. Tested with various custom workloads, AutoDFL reaches an average throughput of over 3000 TPS with a gas reduction of up to 20X.

Problem

• Blockchain Federated Learning
• Performance Optimization
• Security and Privacy

Innovation

• AutoDFL
• zk-Rollups
• Decentralized Federated Learning