Matches in SemOpenAlex for { <https://semopenalex.org/work/W4386565699> ?p ?o ?g. }
Showing items 1 to 79 of
79
with 100 items per page.
- W4386565699 endingPage "714" @default.
- W4386565699 startingPage "706" @default.
- W4386565699 abstract "Adversarial training is considered one of the most effective methods to improve the adversarial robustness of deep neural networks. Despite the success, it still suffers from unsatisfactory performance and overfitting. Considering the intrinsic mechanism of adversarial training, recent studies adopt the idea of curriculum learning to alleviate overfitting. However, this also introduces new issues, that is, lacking the quantitative criterion for attacks' strength and catastrophic forgetting. To mitigate such issues, we propose the self-paced adversarial training (SPAT), which explicitly builds the learning process of adversarial training based on adversarial examples of the whole dataset. Specifically, our model is first trained with easy adversarial examples, and then is continuously enhanced by gradually adding complex adversarial examples. This way strengthens the ability to fit complex adversarial examples while holding in mind easy adversarial samples. To balance adversarial examples between classes, we determine the difficulty of the adversarial examples locally in each class. Notably, this learning paradigm can also be incorporated into other advanced methods for further boosting adversarial robustness. Experimental results show the effectiveness of our proposed model against various attacks on widely-used benchmarks. Especially, on CIFAR100, SPAT provides a boost of 1.7% (relatively 5.4%) in robust accuracy on the PGD10 attack and 3.9% (relatively 7.2%) in natural accuracy for AWP." @default.
- W4386565699 created "2023-09-10" @default.
- W4386565699 creator A5003763794 @default.
- W4386565699 creator A5006340696 @default.
- W4386565699 creator A5051227924 @default.
- W4386565699 creator A5055007304 @default.
- W4386565699 creator A5055826974 @default.
- W4386565699 creator A5062987887 @default.
- W4386565699 date "2023-10-01" @default.
- W4386565699 modified "2023-10-18" @default.
- W4386565699 title "Boosting adversarial robustness via self-paced adversarial training" @default.
- W4386565699 cites W1995137594 @default.
- W4386565699 cites W2296073425 @default.
- W4386565699 cites W2603766943 @default.
- W4386565699 cites W2739553566 @default.
- W4386565699 cites W2948657385 @default.
- W4386565699 cites W3120894151 @default.
- W4386565699 cites W3162368724 @default.
- W4386565699 cites W3206581298 @default.
- W4386565699 cites W4282918038 @default.
- W4386565699 cites W4308347898 @default.
- W4386565699 doi "https://doi.org/10.1016/j.neunet.2023.08.063" @default.
- W4386565699 hasPubMedId "https://pubmed.ncbi.nlm.nih.gov/37729786" @default.
- W4386565699 hasPublicationYear "2023" @default.
- W4386565699 type Work @default.
- W4386565699 citedByCount "0" @default.
- W4386565699 crossrefType "journal-article" @default.
- W4386565699 hasAuthorship W4386565699A5003763794 @default.
- W4386565699 hasAuthorship W4386565699A5006340696 @default.
- W4386565699 hasAuthorship W4386565699A5051227924 @default.
- W4386565699 hasAuthorship W4386565699A5055007304 @default.
- W4386565699 hasAuthorship W4386565699A5055826974 @default.
- W4386565699 hasAuthorship W4386565699A5062987887 @default.
- W4386565699 hasConcept C104317684 @default.
- W4386565699 hasConcept C108583219 @default.
- W4386565699 hasConcept C119857082 @default.
- W4386565699 hasConcept C154945302 @default.
- W4386565699 hasConcept C185592680 @default.
- W4386565699 hasConcept C22019652 @default.
- W4386565699 hasConcept C2984842247 @default.
- W4386565699 hasConcept C37736160 @default.
- W4386565699 hasConcept C41008148 @default.
- W4386565699 hasConcept C46686674 @default.
- W4386565699 hasConcept C50644808 @default.
- W4386565699 hasConcept C55493867 @default.
- W4386565699 hasConcept C63479239 @default.
- W4386565699 hasConceptScore W4386565699C104317684 @default.
- W4386565699 hasConceptScore W4386565699C108583219 @default.
- W4386565699 hasConceptScore W4386565699C119857082 @default.
- W4386565699 hasConceptScore W4386565699C154945302 @default.
- W4386565699 hasConceptScore W4386565699C185592680 @default.
- W4386565699 hasConceptScore W4386565699C22019652 @default.
- W4386565699 hasConceptScore W4386565699C2984842247 @default.
- W4386565699 hasConceptScore W4386565699C37736160 @default.
- W4386565699 hasConceptScore W4386565699C41008148 @default.
- W4386565699 hasConceptScore W4386565699C46686674 @default.
- W4386565699 hasConceptScore W4386565699C50644808 @default.
- W4386565699 hasConceptScore W4386565699C55493867 @default.
- W4386565699 hasConceptScore W4386565699C63479239 @default.
- W4386565699 hasLocation W43865656991 @default.
- W4386565699 hasLocation W43865656992 @default.
- W4386565699 hasOpenAccess W4386565699 @default.
- W4386565699 hasPrimaryLocation W43865656991 @default.
- W4386565699 hasRelatedWork W1487831638 @default.
- W4386565699 hasRelatedWork W2121565117 @default.
- W4386565699 hasRelatedWork W2393964553 @default.
- W4386565699 hasRelatedWork W2738001131 @default.
- W4386565699 hasRelatedWork W2950183588 @default.
- W4386565699 hasRelatedWork W2997056298 @default.
- W4386565699 hasRelatedWork W3080754722 @default.
- W4386565699 hasRelatedWork W3093978547 @default.
- W4386565699 hasRelatedWork W3099765033 @default.
- W4386565699 hasRelatedWork W3203790781 @default.
- W4386565699 hasVolume "167" @default.
- W4386565699 isParatext "false" @default.
- W4386565699 isRetracted "false" @default.
- W4386565699 workType "article" @default.