Unified Breakdown Analysis for Byzantine Robust Gossip

Renaud Gaucher, Aymeric Dieuleveut, Hadrien Hendrikx
Proceedings of the 42nd International Conference on Machine Learning, PMLR 267:18868-18896, 2025.

Abstract

In decentralized machine learning, different devices communicate in a peer-to-peer manner to collaboratively learn from each other’s data. Such approaches are vulnerable to misbehaving (or Byzantine) devices. We introduce F-RG, a general framework for building robust decentralized algorithms with guarantees arising from robust-sum-like aggregation rules F. We then investigate the notion of breakdown point, and show an upper bound on the number of adversaries that decentralized algorithms can tolerate. We introduce a practical robust aggregation rule, coined CS+, such that CS+-RG has a near-optimal breakdown. Other choices of aggregation rules lead to existing algorithms such as ClippedGossip or NNA. We give experimental evidence to validate the effectiveness of CS+-RG and highlight the gap with NNA, in particular against a novel attack tailored to decentralized communications.

Cite this Paper

BibTeX
@InProceedings{pmlr-v267-gaucher25a, title = {Unified Breakdown Analysis for {B}yzantine Robust Gossip}, author = {Gaucher, Renaud and Dieuleveut, Aymeric and Hendrikx, Hadrien}, booktitle = {Proceedings of the 42nd International Conference on Machine Learning}, pages = {18868--18896}, year = {2025}, editor = {Singh, Aarti and Fazel, Maryam and Hsu, Daniel and Lacoste-Julien, Simon and Berkenkamp, Felix and Maharaj, Tegan and Wagstaff, Kiri and Zhu, Jerry}, volume = {267}, series = {Proceedings of Machine Learning Research}, month = {13--19 Jul}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v267/main/assets/gaucher25a/gaucher25a.pdf}, url = {https://proceedings.mlr.press/v267/gaucher25a.html}, abstract = {In decentralized machine learning, different devices communicate in a peer-to-peer manner to collaboratively learn from each other’s data. Such approaches are vulnerable to misbehaving (or Byzantine) devices. We introduce F-RG, a general framework for building robust decentralized algorithms with guarantees arising from robust-sum-like aggregation rules F. We then investigate the notion of breakdown point, and show an upper bound on the number of adversaries that decentralized algorithms can tolerate. We introduce a practical robust aggregation rule, coined CS+, such that CS+-RG has a near-optimal breakdown. Other choices of aggregation rules lead to existing algorithms such as ClippedGossip or NNA. We give experimental evidence to validate the effectiveness of CS+-RG and highlight the gap with NNA, in particular against a novel attack tailored to decentralized communications.} }
Endnote
%0 Conference Paper %T Unified Breakdown Analysis for Byzantine Robust Gossip %A Renaud Gaucher %A Aymeric Dieuleveut %A Hadrien Hendrikx %B Proceedings of the 42nd International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2025 %E Aarti Singh %E Maryam Fazel %E Daniel Hsu %E Simon Lacoste-Julien %E Felix Berkenkamp %E Tegan Maharaj %E Kiri Wagstaff %E Jerry Zhu %F pmlr-v267-gaucher25a %I PMLR %P 18868--18896 %U https://proceedings.mlr.press/v267/gaucher25a.html %V 267 %X In decentralized machine learning, different devices communicate in a peer-to-peer manner to collaboratively learn from each other’s data. Such approaches are vulnerable to misbehaving (or Byzantine) devices. We introduce F-RG, a general framework for building robust decentralized algorithms with guarantees arising from robust-sum-like aggregation rules F. We then investigate the notion of breakdown point, and show an upper bound on the number of adversaries that decentralized algorithms can tolerate. We introduce a practical robust aggregation rule, coined CS+, such that CS+-RG has a near-optimal breakdown. Other choices of aggregation rules lead to existing algorithms such as ClippedGossip or NNA. We give experimental evidence to validate the effectiveness of CS+-RG and highlight the gap with NNA, in particular against a novel attack tailored to decentralized communications.
APA
Gaucher, R., Dieuleveut, A. & Hendrikx, H.. (2025). Unified Breakdown Analysis for Byzantine Robust Gossip. Proceedings of the 42nd International Conference on Machine Learning, in Proceedings of Machine Learning Research 267:18868-18896 Available from https://proceedings.mlr.press/v267/gaucher25a.html.

Related Material