EURO-Online login
- New to EURO? Create an account
- I forgot my username and/or my password.
- Help with cookies
(important for IE8 users)
2110. A framework for modelling view change in Byzantine fault-tolerance network
Invited abstract in session TC-28: Advancements of OR-analytics in statistics, machine learning and data science 6, stream Advancements of OR-analytics in statistics, machine learning and data science.
Tuesday, 12:30-14:00Room: 065 (building: 208)
Authors (first author is the speaker)
1. | Yifei Xie
|
School of Informatics, University of Edinburgh | |
2. | Btissam Er-Rahmadi
|
Knowledge Graph Lab, Huawei Technologies R&D | |
3. | Xiao Chen
|
University of Leicester | |
4. | Tiejun Ma
|
Inforamatics, University of Edinburgh | |
5. | Jane Hillston
|
Informatics, University of Edinburgh |
Abstract
In the rapidly evolving field of machine learning, the increasing complexity of models necessitates larger datasets and greater computational resources. This demand for scalability and enhanced computational capabilities has driven the study of distributed optimization to the forefront. Byzantine fault-tolerance (BFT) consensus is a fundamental building block of distributed systems. BFT is the property of a system that can resist the class of failures which can continue operating even if some of the nodes fail or act maliciously, thanks to the capacity to prevent the no more than one third adversaries from gaining a consensus and the mechanism of view changes for replacing faulty nodes.
Our research introduces a novel mathematical programming framework aimed at optimizing the performance of BFT algorithms. Our approach is twofold: firstly, we optimize node network distribution strategies to maximize throughput; secondly, we investigate mechanisms for selecting backup nodes to replace the leader during view change phases when failure occur. To validate our framework, we conducted experiments on a real-world testbed. The results demonstrate significant improvements in BFT algorithm performance, showcasing our method's potential in optimizing throughput and ensuring robust fault tolerance during view changes. This study contributes to the development of more resilient and efficient distributed systems, supporting the growing demands of machine learning applications.
Keywords
- Machine Learning
- Optimization Modeling
- Computer Science/Applications
Status: accepted
Back to the list of papers