Averaging three observation filters (random noise, autoencoder, PCA) before action selection substantially improves a Highway-env DQN's reward and collision rate under FGSM attacks.
Online Robustness Training for Deep Reinforcement Learning
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
abstract
In deep reinforcement learning (RL), adversarial attacks can trick an agent into unwanted states and disrupt training. We propose a system called Robust Student-DQN (RS-DQN), which permits online robustness training alongside Q networks, while preserving competitive performance. We show that RS-DQN can be combined with (i) state-of-the-art adversarial training and (ii) provably robust training to obtain an agent that is resilient to strong attacks during training and evaluation.
citation-role summary
background 1
citation-polarity summary
fields
cs.LG 1years
2025 1verdicts
CONDITIONAL 1roles
background 1polarities
unclear 1representative citing papers
citing papers explorer
-
Advancing Robustness in Deep Reinforcement Learning with an Ensemble Defense Approach
Averaging three observation filters (random noise, autoencoder, PCA) before action selection substantially improves a Highway-env DQN's reward and collision rate under FGSM attacks.