The Normalized attack breaks Byzantine-robust federated reinforcement learning by maximizing angular deviation in policy updates, and a group-based ensemble with majority vote or geometric median provides a narrower-than-claimed provable defense.
A little is enough: Circumvent- ing defenses for distributed learning
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
citation-role summary
background 1
citation-polarity summary
fields
cs.CR 1years
2025 1verdicts
CONDITIONAL 1roles
background 1polarities
unclear 1representative citing papers
citing papers explorer
-
Provably Robust Federated Reinforcement Learning
The Normalized attack breaks Byzantine-robust federated reinforcement learning by maximizing angular deviation in policy updates, and a group-based ensemble with majority vote or geometric median provides a narrower-than-claimed provable defense.