Backdoors in deep RL agents can be planted by compromising a training component or by editing pretrained weights with no training data, matching training-time attack success on six Atari games.
Execute Order 66: Targeted Data Poisoning for Reinforcement Learning
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
abstract
Data poisoning for reinforcement learning has historically focused on general performance degradation, and targeted attacks have been successful via perturbations that involve control of the victim's policy and rewards. We introduce an insidious poisoning attack for reinforcement learning which causes agent misbehavior only at specific target states - all while minimally modifying a small fraction of training observations without assuming any control over policy or reward. We accomplish this by adapting a recent technique, gradient alignment, to reinforcement learning. We test our method and demonstrate success in two Atari games of varying difficulty.
fields
cs.LG 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Beyond Training-time Poisoning: Component-level and Post-training Backdoors in Deep Reinforcement Learning
Backdoors in deep RL agents can be planted by compromising a training component or by editing pretrained weights with no training data, matching training-time attack success on six Atari games.