Matches in SemOpenAlex for { <https://semopenalex.org/work/W4287331367> ?p ?o ?g. }
Showing items 1 to 62 of
62
with 100 items per page.
- W4287331367 abstract "High cost of training time caused by multi-step adversarial example generation is a major challenge in adversarial training. Previous methods try to reduce the computational burden of adversarial training using single-step adversarial example generation schemes, which can effectively improve the efficiency but also introduce the problem of catastrophic overfitting, where the robust accuracy against Fast Gradient Sign Method (FGSM) can achieve nearby 100% whereas the robust accuracy against Projected Gradient Descent (PGD) suddenly drops to 0% over a single epoch. To address this problem, we propose a novel Fast Gradient Sign Method with PGD Regularization (FGSMPR) to boost the efficiency of adversarial training without catastrophic overfitting. Our core idea is that single-step adversarial training can not learn robust internal representations of FGSM and PGD adversarial examples. Therefore, we design a PGD regularization term to encourage similar embeddings of FGSM and PGD adversarial examples. The experiments demonstrate that our proposed method can train a robust deep network for L$_infty$-perturbations with FGSM adversarial training and reduce the gap to multi-step adversarial training." @default.
- W4287331367 created "2022-07-25" @default.
- W4287331367 creator A5028630686 @default.
- W4287331367 creator A5058120371 @default.
- W4287331367 creator A5061126226 @default.
- W4287331367 creator A5079784316 @default.
- W4287331367 date "2021-02-05" @default.
- W4287331367 modified "2023-09-28" @default.
- W4287331367 title "Robust Single-step Adversarial Training with Regularizer" @default.
- W4287331367 hasPublicationYear "2021" @default.
- W4287331367 type Work @default.
- W4287331367 citedByCount "0" @default.
- W4287331367 crossrefType "posted-content" @default.
- W4287331367 hasAuthorship W4287331367A5028630686 @default.
- W4287331367 hasAuthorship W4287331367A5058120371 @default.
- W4287331367 hasAuthorship W4287331367A5061126226 @default.
- W4287331367 hasAuthorship W4287331367A5079784316 @default.
- W4287331367 hasBestOaLocation W42873313671 @default.
- W4287331367 hasConcept C11413529 @default.
- W4287331367 hasConcept C119857082 @default.
- W4287331367 hasConcept C126255220 @default.
- W4287331367 hasConcept C153258448 @default.
- W4287331367 hasConcept C153294291 @default.
- W4287331367 hasConcept C154945302 @default.
- W4287331367 hasConcept C205649164 @default.
- W4287331367 hasConcept C22019652 @default.
- W4287331367 hasConcept C2776135515 @default.
- W4287331367 hasConcept C2777211547 @default.
- W4287331367 hasConcept C33923547 @default.
- W4287331367 hasConcept C37736160 @default.
- W4287331367 hasConcept C41008148 @default.
- W4287331367 hasConcept C50644808 @default.
- W4287331367 hasConceptScore W4287331367C11413529 @default.
- W4287331367 hasConceptScore W4287331367C119857082 @default.
- W4287331367 hasConceptScore W4287331367C126255220 @default.
- W4287331367 hasConceptScore W4287331367C153258448 @default.
- W4287331367 hasConceptScore W4287331367C153294291 @default.
- W4287331367 hasConceptScore W4287331367C154945302 @default.
- W4287331367 hasConceptScore W4287331367C205649164 @default.
- W4287331367 hasConceptScore W4287331367C22019652 @default.
- W4287331367 hasConceptScore W4287331367C2776135515 @default.
- W4287331367 hasConceptScore W4287331367C2777211547 @default.
- W4287331367 hasConceptScore W4287331367C33923547 @default.
- W4287331367 hasConceptScore W4287331367C37736160 @default.
- W4287331367 hasConceptScore W4287331367C41008148 @default.
- W4287331367 hasConceptScore W4287331367C50644808 @default.
- W4287331367 hasLocation W42873313671 @default.
- W4287331367 hasOpenAccess W4287331367 @default.
- W4287331367 hasPrimaryLocation W42873313671 @default.
- W4287331367 hasRelatedWork W11023528 @default.
- W4287331367 hasRelatedWork W11033548 @default.
- W4287331367 hasRelatedWork W11152069 @default.
- W4287331367 hasRelatedWork W11562254 @default.
- W4287331367 hasRelatedWork W11720838 @default.
- W4287331367 hasRelatedWork W12166850 @default.
- W4287331367 hasRelatedWork W14688374 @default.
- W4287331367 hasRelatedWork W2700343 @default.
- W4287331367 hasRelatedWork W3422034 @default.
- W4287331367 hasRelatedWork W9035903 @default.
- W4287331367 isParatext "false" @default.
- W4287331367 isRetracted "false" @default.
- W4287331367 workType "article" @default.