BagChain: A Dual-Functional Blockchain Leveraging Bagging-Based Distributed Machine Learning
- 23 October 2025
- journal article
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Mobile Computing
- Vol. 25 (3) , 4206-4222
Abstract
Exploiting on-device data and computing power for machine learning at the network edge is challenged by constrained device resources, privacy requirements, and local data heterogeneity. To address the above gap, this work proposes a dual-functional blockchain framework named BagChain for bagging-based decentralized ML. BagChain integrates blockchain with distributed ML by replacing the computationally costly hash computing in proof-of-work with ML model training and validation, and does not rely on any trusted central servers. Individual miners in BagChain train base models by using their local computing resources and private data and further aggregate these base models, which could be very weak, into strong ensemble models. More specifically, we design a three-layer blockchain structure and associated generation and validation mechanisms to enable distributed ML among uncoordinated miners without revealing raw data. To reduce computational waste due to blockchain forking, we further propose the cross fork sharing mechanism for practical networks with lengthy delay and limited bandwidth. Extensive experiments illustrate the superiority and efficacy of BagChain when handling various ML tasks on both independently and identically distributed (IID) and non-IID datasets. BagChain remains robust and effective even when facing resource-constrained mobile devices, heterogeneous private user data, and limited network connectivity.Keywords
All Related Versions
Funding Information
- National Key R&D Program of China (2022YFB2902205)
- National Natural Science Foundation of China (62471137)
- National Natural Science Foundation of China (72231002)
- National Natural Science Foundation of China (U22B2006)
- Natural Science Foundation on Frontier Leading Technology Basic Research Project of Jiangsu (BK20222001)
- Natural Science Foundation on Frontier Leading Technology Basic Research Project of Jiangsu (BK20212001)
- Key Technologies R&D Program of Jiangsu (BE2022068-3)
- Jiangsu Province Major Science and Technology Project (BG2024005)
- Open Research Fund of Key Lab of Broadband Wireless Communication and Sensor Network Technology
- Nanjing University of Posts and Telecommunications
- Ministry of Education (JZNY202313)
- Science and Technology Major Project of Nanjing (202405020)
- Taihu Lake Innovation Fund for the School of Future Technology of Southeast University
- Fundamental Research Funds for the Central Universities (2242022K60002)
- Young Elite Scientists Sponsorship Program by CAST (2023QNRC001)
- Huawei Cooperation Project (FA2019051081-2021-01)
This publication has 47 references indexed in Scilit:
- DePoL: Assuring training integrity in collaborative learning via decentralized verificationJournal of Parallel and Distributed Computing, 2025
- Provably Secure Blockchain Protocols from Distributed Proof-of-Deep-Learning2023
- A collaboration strategy in the mining pool for proof-of-neural-architecture consensusBlockchain: Research and Applications, 2022
- Proof of Learning (PoLe): Empowering neural network training with consensus building on blockchainsComputer Networks, 2021
- Federated learning on non-IID data: A surveyNeurocomputing, 2021
- FedCoin: A Peer-to-Peer Payment System for Federated Learning2020
- DLchain: Blockchain with Deep Learning as Proof-of-Useful-Work2020
- Distributed machine learning in networks by consensusNeurocomputing, 2014
- Gradient-based learning applied to document recognitionProceedings of the IEEE, 1998
- Bagging predictorsMachine Learning, 1996