Skip to content
InfoResearchPeer-reviewed

On the Relevance of Byzantine Robust Optimization Against Data Poisoning

Published
Record updated
View JSON

Summary

The paper examines whether Byzantine machine learning, where distributed workers may deviate arbitrarily from the algorithm, is relevant to data poisoning, a weaker threat model where only local datasets are corrupted. The authors prove that Byzantine-robust schemes yield optimal solutions under both fully-poisonous and partially-poisonous local data, and that fully-poisonous workers are more harmful when local data is heterogeneous.