Matches in SemOpenAlex for { <https://semopenalex.org/work/W2976345833> ?p ?o ?g. }
- W2976345833 abstract "It has been widely recognized that adversarial examples can be easily crafted to fool deep networks, which mainly root from the locally non-linear behavior nearby input examples. Applying mixup in training provides an effective mechanism to improve generalization performance and model robustness against adversarial perturbations, which introduces the globally linear behavior in-between training examples. However, in previous work, the mixup-trained models only passively defend adversarial attacks in inference by directly classifying the inputs, where the induced global linearity is not well exploited. Namely, since the locality of the adversarial perturbations, it would be more efficient to actively break the locality via the globality of the model predictions. Inspired by simple geometric intuition, we develop an inference principle, named mixup inference (MI), for mixup-trained models. MI mixups the input with other random clean samples, which can shrink and transfer the equivalent perturbation if the input is adversarial. Our experiments on CIFAR-10 and CIFAR-100 demonstrate that MI can further improve the adversarial robustness for the models trained by mixup and its variants." @default.
- W2976345833 created "2019-10-03" @default.
- W2976345833 creator A5003713785 @default.
- W2976345833 creator A5029335824 @default.
- W2976345833 creator A5031863731 @default.
- W2976345833 date "2019-09-25" @default.
- W2976345833 modified "2023-09-25" @default.
- W2976345833 title "Mixup Inference: Better Exploiting Mixup to Defend Adversarial Attacks" @default.
- W2976345833 cites W1932198206 @default.
- W2976345833 cites W1980287119 @default.
- W2976345833 cites W2156909104 @default.
- W2976345833 cites W2302255633 @default.
- W2976345833 cites W2773726006 @default.
- W2976345833 cites W2774644650 @default.
- W2976345833 cites W2783482415 @default.
- W2976345833 cites W2795727551 @default.
- W2976345833 cites W2808678362 @default.
- W2976345833 cites W2891021639 @default.
- W2976345833 cites W2913848079 @default.
- W2976345833 cites W2921087533 @default.
- W2976345833 cites W2921320475 @default.
- W2976345833 cites W2949103145 @default.
- W2976345833 cites W2952411227 @default.
- W2976345833 cites W2953355478 @default.
- W2976345833 cites W2954978443 @default.
- W2976345833 cites W2962729158 @default.
- W2976345833 cites W2963001136 @default.
- W2976345833 cites W2963003451 @default.
- W2976345833 cites W2963031676 @default.
- W2976345833 cites W2963143631 @default.
- W2976345833 cites W2963207607 @default.
- W2976345833 cites W2963399829 @default.
- W2976345833 cites W2963467071 @default.
- W2976345833 cites W2963539306 @default.
- W2976345833 cites W2963542245 @default.
- W2976345833 cites W2963564844 @default.
- W2976345833 cites W2963777966 @default.
- W2976345833 cites W2963849784 @default.
- W2976345833 cites W2963920068 @default.
- W2976345833 cites W2964153729 @default.
- W2976345833 cites W2964253222 @default.
- W2976345833 cites W2964283260 @default.
- W2976345833 cites W2964301649 @default.
- W2976345833 cites W2964328535 @default.
- W2976345833 cites W2965595599 @default.
- W2976345833 cites W2970317235 @default.
- W2976345833 cites W2971970905 @default.
- W2976345833 cites W2978426779 @default.
- W2976345833 cites W3118608800 @default.
- W2976345833 cites W3137695714 @default.
- W2976345833 cites W3146803896 @default.
- W2976345833 hasPublicationYear "2019" @default.
- W2976345833 type Work @default.
- W2976345833 sameAs 2976345833 @default.
- W2976345833 citedByCount "7" @default.
- W2976345833 countsByYear W29763458332020 @default.
- W2976345833 countsByYear W29763458332021 @default.
- W2976345833 countsByYear W29763458332022 @default.
- W2976345833 crossrefType "posted-content" @default.
- W2976345833 hasAuthorship W2976345833A5003713785 @default.
- W2976345833 hasAuthorship W2976345833A5029335824 @default.
- W2976345833 hasAuthorship W2976345833A5031863731 @default.
- W2976345833 hasConcept C104317684 @default.
- W2976345833 hasConcept C111472728 @default.
- W2976345833 hasConcept C119857082 @default.
- W2976345833 hasConcept C132010649 @default.
- W2976345833 hasConcept C138885662 @default.
- W2976345833 hasConcept C154945302 @default.
- W2976345833 hasConcept C185592680 @default.
- W2976345833 hasConcept C2776214188 @default.
- W2976345833 hasConcept C2779808786 @default.
- W2976345833 hasConcept C37736160 @default.
- W2976345833 hasConcept C41008148 @default.
- W2976345833 hasConcept C41895202 @default.
- W2976345833 hasConcept C55493867 @default.
- W2976345833 hasConcept C63479239 @default.
- W2976345833 hasConceptScore W2976345833C104317684 @default.
- W2976345833 hasConceptScore W2976345833C111472728 @default.
- W2976345833 hasConceptScore W2976345833C119857082 @default.
- W2976345833 hasConceptScore W2976345833C132010649 @default.
- W2976345833 hasConceptScore W2976345833C138885662 @default.
- W2976345833 hasConceptScore W2976345833C154945302 @default.
- W2976345833 hasConceptScore W2976345833C185592680 @default.
- W2976345833 hasConceptScore W2976345833C2776214188 @default.
- W2976345833 hasConceptScore W2976345833C2779808786 @default.
- W2976345833 hasConceptScore W2976345833C37736160 @default.
- W2976345833 hasConceptScore W2976345833C41008148 @default.
- W2976345833 hasConceptScore W2976345833C41895202 @default.
- W2976345833 hasConceptScore W2976345833C55493867 @default.
- W2976345833 hasConceptScore W2976345833C63479239 @default.
- W2976345833 hasLocation W29763458331 @default.
- W2976345833 hasOpenAccess W2976345833 @default.
- W2976345833 hasPrimaryLocation W29763458331 @default.
- W2976345833 hasRelatedWork W2194775991 @default.
- W2976345833 hasRelatedWork W2765407302 @default.
- W2976345833 hasRelatedWork W2777353073 @default.
- W2976345833 hasRelatedWork W2781758978 @default.
- W2976345833 hasRelatedWork W2804545960 @default.
- W2976345833 hasRelatedWork W2890591829 @default.
- W2976345833 hasRelatedWork W2950942529 @default.