FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W3136758972> ?p ?o ?g. }

• W3136758972 abstract "Deep Neural Networks (DNNs) have achieved remarkable performance on a variety of applications but are extremely vulnerable to adversarial perturbation. To address this issue, various defense methods have been proposed to enhance model robustness. Unfortunately, the most representative and promising methods, such as adversarial training and its variants, usually degrade model accuracy on benign samples, limiting practical utility. This indicates that it is difficult to extract both robust and accurate features using a single network under certain conditions, such as limited training data, resulting in a trade-off between accuracy and robustness. To tackle this problem, we propose an Adversarial Feature Stacking (AFS) model that can jointly take advantage of features with varied levels of robustness and accuracy, thus significantly alleviating the aforementioned trade-off. Specifically, we adopt multiple networks adversarially trained with different perturbation budgets to extract either more robust features or more accurate features. These features are then fused by a learnable merger to give final predictions. We evaluate the AFS model on CIFAR-10 and CIFAR-100 datasets with strong adaptive attack methods, which significantly advances the state-of-the-art in terms of the trade-off. Without extra training data, the AFS model achieves a benign accuracy improvement of 6% on CIFAR-10 and 9% on CIFAR-100 with comparable or even stronger robustness than the state-of-the-art adversarial training methods. This work demonstrates the feasibility to obtain both accurate and robust models under the circumstances of limited training data." @default.
• W3136758972 created "2021-03-29" @default.
• W3136758972 creator A5018476895 @default.
• W3136758972 creator A5037328964 @default.
• W3136758972 creator A5037705126 @default.
• W3136758972 date "2021-03-24" @default.
• W3136758972 modified "2023-09-27" @default.
• W3136758972 title "Adversarial Feature Stacking for Accurate and Robust Predictions." @default.
• W3136758972 cites W2194775991 @default.
• W3136758972 cites W2395611524 @default.
• W3136758972 cites W2401231614 @default.
• W3136758972 cites W2460937040 @default.
• W3136758972 cites W2503523779 @default.
• W3136758972 cites W2543927648 @default.
• W3136758972 cites W2607219512 @default.
• W3136758972 cites W2765233338 @default.
• W3136758972 cites W2766462876 @default.
• W3136758972 cites W2774018344 @default.
• W3136758972 cites W2774644650 @default.
• W3136758972 cites W2786118190 @default.
• W3136758972 cites W2786163515 @default.
• W3136758972 cites W2787708942 @default.
• W3136758972 cites W2787733970 @default.
• W3136758972 cites W2803392236 @default.
• W3136758972 cites W2874797877 @default.
• W3136758972 cites W2897355816 @default.
• W3136758972 cites W2901213450 @default.
• W3136758972 cites W2913266441 @default.
• W3136758972 cites W2919115771 @default.
• W3136758972 cites W2949362401 @default.
• W3136758972 cites W2962872506 @default.
• W3136758972 cites W2963207607 @default.
• W3136758972 cites W2963341956 @default.
• W3136758972 cites W2963403868 @default.
• W3136758972 cites W2963496101 @default.
• W3136758972 cites W2963744840 @default.
• W3136758972 cites W2964082701 @default.
• W3136758972 cites W2964153729 @default.
• W3136758972 cites W2964253222 @default.
• W3136758972 cites W2970088379 @default.
• W3136758972 cites W2970115835 @default.
• W3136758972 cites W2970680991 @default.
• W3136758972 cites W2995245581 @default.
• W3136758972 cites W3006834354 @default.
• W3136758972 cites W3006935033 @default.
• W3136758972 cites W3007305010 @default.
• W3136758972 cites W3009542902 @default.
• W3136758972 cites W3030163527 @default.
• W3136758972 cites W3034537217 @default.
• W3136758972 cites W3035032188 @default.
• W3136758972 cites W3037182574 @default.
• W3136758972 cites W3092171228 @default.
• W3136758972 cites W3101700548 @default.
• W3136758972 cites W3103385169 @default.
• W3136758972 cites W3103557498 @default.
• W3136758972 cites W3106250896 @default.
• W3136758972 cites W3107235539 @default.
• W3136758972 cites W3109453310 @default.
• W3136758972 cites W3120915204 @default.
• W3136758972 cites W3125614726 @default.
• W3136758972 hasPublicationYear "2021" @default.
• W3136758972 type Work @default.
• W3136758972 sameAs 3136758972 @default.
• W3136758972 citedByCount "0" @default.
• W3136758972 crossrefType "posted-content" @default.
• W3136758972 hasAuthorship W3136758972A5018476895 @default.
• W3136758972 hasAuthorship W3136758972A5037328964 @default.
• W3136758972 hasAuthorship W3136758972A5037705126 @default.
• W3136758972 hasConcept C104317684 @default.
• W3136758972 hasConcept C119857082 @default.
• W3136758972 hasConcept C121332964 @default.
• W3136758972 hasConcept C124101348 @default.
• W3136758972 hasConcept C127413603 @default.
• W3136758972 hasConcept C153180895 @default.
• W3136758972 hasConcept C154945302 @default.
• W3136758972 hasConcept C185592680 @default.
• W3136758972 hasConcept C188198153 @default.
• W3136758972 hasConcept C2984842247 @default.
• W3136758972 hasConcept C33347731 @default.
• W3136758972 hasConcept C37736160 @default.
• W3136758972 hasConcept C41008148 @default.
• W3136758972 hasConcept C46141821 @default.
• W3136758972 hasConcept C50644808 @default.
• W3136758972 hasConcept C51632099 @default.
• W3136758972 hasConcept C55493867 @default.
• W3136758972 hasConcept C63479239 @default.
• W3136758972 hasConcept C78519656 @default.
• W3136758972 hasConceptScore W3136758972C104317684 @default.
• W3136758972 hasConceptScore W3136758972C119857082 @default.
• W3136758972 hasConceptScore W3136758972C121332964 @default.
• W3136758972 hasConceptScore W3136758972C124101348 @default.
• W3136758972 hasConceptScore W3136758972C127413603 @default.
• W3136758972 hasConceptScore W3136758972C153180895 @default.
• W3136758972 hasConceptScore W3136758972C154945302 @default.
• W3136758972 hasConceptScore W3136758972C185592680 @default.
• W3136758972 hasConceptScore W3136758972C188198153 @default.
• W3136758972 hasConceptScore W3136758972C2984842247 @default.
• W3136758972 hasConceptScore W3136758972C33347731 @default.
• W3136758972 hasConceptScore W3136758972C37736160 @default.
• W3136758972 hasConceptScore W3136758972C41008148 @default.

• next