Matches in SemOpenAlex for { <https://semopenalex.org/work/W3128448405> ?p ?o ?g. }
Showing items 1 to 90 of
90
with 100 items per page.
- W3128448405 abstract "Adversarial training has been proven to be an effective technique for improving the adversarial robustness of models. However, there seems to be an inherent trade-off between optimizing the model for accuracy and robustness. To this end, we propose Adversarial Concurrent Training (ACT), which employs adversarial training in a collaborative learning framework whereby we train a robust model in conjunction with a natural model in a minimax game. ACT encourages the two models to align their feature space by using the task-specific decision boundaries and explore the input space more broadly. Furthermore, the natural model acts as a regularizer, enforcing priors on features that the robust model should learn. Our analyses on the behavior of the models show that ACT leads to a robust model with lower model complexity, higher information compression in the learned representations, and high posterior entropy solutions indicative of convergence to a flatter minima. We demonstrate the effectiveness of the proposed approach across different datasets and network architectures. On ImageNet, ACT achieves 68.20% standard accuracy and 44.29% robustness accuracy under a 100-iteration untargeted attack, improving upon the standard adversarial training method's 65.70% standard accuracy and 42.36% robustness." @default.
- W3128448405 created "2021-02-15" @default.
- W3128448405 creator A5027547386 @default.
- W3128448405 creator A5028650455 @default.
- W3128448405 creator A5047435371 @default.
- W3128448405 date "2020-08-16" @default.
- W3128448405 modified "2023-09-26" @default.
- W3128448405 title "Adversarial Concurrent Training: Optimizing Robustness and Accuracy Trade-off of Deep Neural Networks." @default.
- W3128448405 cites W2736506089 @default.
- W3128448405 cites W2786118190 @default.
- W3128448405 cites W2791122243 @default.
- W3128448405 cites W2951163624 @default.
- W3128448405 cites W2963735582 @default.
- W3128448405 cites W2963857521 @default.
- W3128448405 cites W2970317235 @default.
- W3128448405 hasPublicationYear "2020" @default.
- W3128448405 type Work @default.
- W3128448405 sameAs 3128448405 @default.
- W3128448405 citedByCount "0" @default.
- W3128448405 crossrefType "proceedings-article" @default.
- W3128448405 hasAuthorship W3128448405A5027547386 @default.
- W3128448405 hasAuthorship W3128448405A5028650455 @default.
- W3128448405 hasAuthorship W3128448405A5047435371 @default.
- W3128448405 hasConcept C104317684 @default.
- W3128448405 hasConcept C106301342 @default.
- W3128448405 hasConcept C107673813 @default.
- W3128448405 hasConcept C119857082 @default.
- W3128448405 hasConcept C121332964 @default.
- W3128448405 hasConcept C126255220 @default.
- W3128448405 hasConcept C134306372 @default.
- W3128448405 hasConcept C149728462 @default.
- W3128448405 hasConcept C154945302 @default.
- W3128448405 hasConcept C177769412 @default.
- W3128448405 hasConcept C185592680 @default.
- W3128448405 hasConcept C186633575 @default.
- W3128448405 hasConcept C2984842247 @default.
- W3128448405 hasConcept C33923547 @default.
- W3128448405 hasConcept C37736160 @default.
- W3128448405 hasConcept C41008148 @default.
- W3128448405 hasConcept C50644808 @default.
- W3128448405 hasConcept C55493867 @default.
- W3128448405 hasConcept C62520636 @default.
- W3128448405 hasConcept C63479239 @default.
- W3128448405 hasConceptScore W3128448405C104317684 @default.
- W3128448405 hasConceptScore W3128448405C106301342 @default.
- W3128448405 hasConceptScore W3128448405C107673813 @default.
- W3128448405 hasConceptScore W3128448405C119857082 @default.
- W3128448405 hasConceptScore W3128448405C121332964 @default.
- W3128448405 hasConceptScore W3128448405C126255220 @default.
- W3128448405 hasConceptScore W3128448405C134306372 @default.
- W3128448405 hasConceptScore W3128448405C149728462 @default.
- W3128448405 hasConceptScore W3128448405C154945302 @default.
- W3128448405 hasConceptScore W3128448405C177769412 @default.
- W3128448405 hasConceptScore W3128448405C185592680 @default.
- W3128448405 hasConceptScore W3128448405C186633575 @default.
- W3128448405 hasConceptScore W3128448405C2984842247 @default.
- W3128448405 hasConceptScore W3128448405C33923547 @default.
- W3128448405 hasConceptScore W3128448405C37736160 @default.
- W3128448405 hasConceptScore W3128448405C41008148 @default.
- W3128448405 hasConceptScore W3128448405C50644808 @default.
- W3128448405 hasConceptScore W3128448405C55493867 @default.
- W3128448405 hasConceptScore W3128448405C62520636 @default.
- W3128448405 hasConceptScore W3128448405C63479239 @default.
- W3128448405 hasLocation W31284484051 @default.
- W3128448405 hasOpenAccess W3128448405 @default.
- W3128448405 hasPrimaryLocation W31284484051 @default.
- W3128448405 hasRelatedWork W2894793033 @default.
- W3128448405 hasRelatedWork W2944797015 @default.
- W3128448405 hasRelatedWork W2950630935 @default.
- W3128448405 hasRelatedWork W2952718061 @default.
- W3128448405 hasRelatedWork W2953355478 @default.
- W3128448405 hasRelatedWork W2979999556 @default.
- W3128448405 hasRelatedWork W2981817257 @default.
- W3128448405 hasRelatedWork W2981924372 @default.
- W3128448405 hasRelatedWork W2989994511 @default.
- W3128448405 hasRelatedWork W3028763319 @default.
- W3128448405 hasRelatedWork W3032971194 @default.
- W3128448405 hasRelatedWork W3049579477 @default.
- W3128448405 hasRelatedWork W3089774673 @default.
- W3128448405 hasRelatedWork W3093813382 @default.
- W3128448405 hasRelatedWork W3094319733 @default.
- W3128448405 hasRelatedWork W3116804412 @default.
- W3128448405 hasRelatedWork W3125045752 @default.
- W3128448405 hasRelatedWork W3135151133 @default.
- W3128448405 hasRelatedWork W3169425223 @default.
- W3128448405 hasRelatedWork W3175117184 @default.
- W3128448405 isParatext "false" @default.
- W3128448405 isRetracted "false" @default.
- W3128448405 magId "3128448405" @default.
- W3128448405 workType "article" @default.