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.
Comment on "Relation between scattering amplitude and Bethe-Salpeter wave function in quantum field theory"
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
abstract
We invalidate the arguments given in [T.Yamazaki and Y.Kuramashi, Phys. Rev. D96, 114511 (2017)] over the HAL QCD method for hadron-hadron interactions on the lattice. We also pose questions on the practical usefulness of the method proposed in this reference.
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.