Single deep MoE layers improve PGD/AutoPGD robust accuracy of adversarially trained ResNets on CIFAR-100, and routing collapse under switch loss produces individual experts that are more robust than the full MoE.
Switch transformers: Scaling to trillion parameter models with sim- ple and efficient sparsity.The Journal of Machine Learning Research, 2022
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.CV 1years
2025 1verdicts
CONDITIONAL 1roles
background 1polarities
unclear 1representative citing papers
citing papers explorer
-
Robust Experts: the Effect of Adversarial Training on CNNs with Sparse Mixture-of-Experts Layers
Single deep MoE layers improve PGD/AutoPGD robust accuracy of adversarially trained ResNets on CIFAR-100, and routing collapse under switch loss produces individual experts that are more robust than the full MoE.