Computer Science > Distributed, Parallel, and Cluster Computing
[Submitted on 31 May 2024]
Title:Wait or Not to Wait: Evaluating Trade-Offs between Speed and Precision in Blockchain-based Federated Aggregation
View PDF HTML (experimental)Abstract:This paper presents a fully coupled blockchain-assisted federated learning architecture that effectively eliminates single points of failure by decentralizing both the training and aggregation tasks across all participants. Our proposed system offers a high degree of flexibility, allowing participants to select shared models and customize the aggregation for local needs, thereby optimizing system performance, including accurate inference results. Notably, the integration of blockchain technology in our work is to promote a trustless environment, ensuring transparency and non-repudiation among participants when abnormalities are detected. To validate the effectiveness, we conducted real-world federated learning deployments on a private Ethereum platform, using two different models, ranging from simple to complex neural networks. The experimental results indicate comparable inference accuracy between centralized and decentralized federated learning settings. Furthermore, our findings indicate that asynchronous aggregation is a feasible option for simple learning models. However, complex learning models require greater training model involvement in the aggregation to achieve high model quality, instead of asynchronous aggregation. With the implementation of asynchronous aggregation and the flexibility to select models, participants anticipate decreased aggregation time in each communication round, while experiencing minimal accuracy trade-off.
References & Citations
Bibliographic and Citation Tools
Bibliographic Explorer (What is the Explorer?)
Litmaps (What is Litmaps?)
scite Smart Citations (What are Smart Citations?)
Code, Data and Media Associated with this Article
CatalyzeX Code Finder for Papers (What is CatalyzeX?)
DagsHub (What is DagsHub?)
Gotit.pub (What is GotitPub?)
Papers with Code (What is Papers with Code?)
ScienceCast (What is ScienceCast?)
Demos
Recommenders and Search Tools
Influence Flower (What are Influence Flowers?)
Connected Papers (What is Connected Papers?)
CORE Recommender (What is CORE?)
arXivLabs: experimental projects with community collaborators
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.