aiqbench analysis
This research presents CESAR, a protocol that successfully reconciles the long-standing conflict between communication efficiency via sparsification and privacy protection in decentralized learning networks. By coordinating masks over parameter intersections rather than relying on a central server, it achieves significant reductions in data exchange while maintaining model accuracy against various adversarial threats. The work is critical for scaling federated AI systems where bandwidth constraints and strict privacy requirements previously forced impossible trade-offs between performance and security.
Communication-Efficient Secure Aggregation in Decentralized Learning
arXiv:2405.07708v3 Announce Type: replace Abstract: Decentralized learning (DL) enables participants to collaboratively train models without a central server, yet it faces significant scalability challenges that demand sparsification to reduce the prohibitive communication costs of peer-to-peer exchange. While secure aggregation effectively mitigates privacy risks in standard settings, it has remained fundamentally incompatible with sparsification in decentralized networks due to the mismatch of indices across local updates, forcing a trade-off between communication efficiency and privacy. This paper introduces CESAR, a novel protocol that resolves this incompatibility by integrating secure aggregation and sparsification to provide provable defense against honest-but-curious and colluding adversaries. By coordinating masks over parameter intersections, CESAR supports node dropouts and robust privacy without central aggregation. Empirical evaluations on models up to 124 million parameters demonstrate that CESAR matches the accuracy of non-private baselines while cutting total data exchange by 66.7 % compared to a standard full-parameter decentralized protocol (D-PSGD). With TopK sparsification on IID data, CESAR even exceeds by 0.3 % the accuracy achieved by D-PSGD with sparsification. Collectively, these results establish CESAR as the first decentralized protocol to achieve both privacy and communication efficiency through secure aggregation in DL.
Abstract:Decentralized learning (DL) enables participants to collaboratively train models without a central server, yet it faces significant scalability challenges that demand sparsification to reduce the prohibitive communication costs of peer-to-peer exchange. While secure aggregation effectively mitigates privacy risks in standard settings, it has remained fundamentally incompatible with sparsification in decentralized networks due to the mismatch of indices across local updates, forcing a trade-off between communication efficiency and privacy. This paper introduces CESAR, a novel protocol that resolves this incompatibility by integrating secure aggregation and sparsification to provide provable defense against honest-but-curious and colluding adversaries. By coordinating masks over parameter intersections, CESAR supports node dropouts and robust privacy without central aggregation. Empirical evaluations on models up to 124 million parameters demonstrate that CESAR matches the accuracy of non-private baselines while cutting total data exchange by 66.7 % compared to a standard full-parameter decentralized protocol (D-PSGD). With TopK sparsification on IID data, CESAR even exceeds by 0.3 % the accuracy achieved by D-PSGD with sparsification. Collectively, these results establish CESAR as the first decentralized protocol to achieve both privacy and communication efficiency through secure aggregation in DL.
Machine Learning (cs.LG)
arXiv:2405.07708 [cs.LG]
(or arXiv:2405.07708v3 [cs.LG] for this version)
https://doi.org/10.48550/arXiv.2405.07708 Focus to learn more arXiv-issued DOI via DataCite
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.
Full text under the publisher's open licence · 277 words
Related coverage
Secure Aggregation with Top-K Sparsification in Decentralized Federated Learning
arXiv · Jun 10, 2026
TITAN-FedAnil+: Trust-Based Adaptive Blockchain Federated Learning for Resource-Constrained Intelligent Enterprises
arXiv · Jun 5, 2026
Optimizing Byzantine Node Placement in Decentralized Federated Learning
arXiv · Sep 2, 2026
Hierarchical Decentralized Multi-Agent Coordination with Privacy-Preserving Knowledge Sharing: Extending AgentNet for Scalable Autonomous Systems
arXiv · Oct 7, 2026