FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W4308347898> ?p ?o ?g. }

• W4308347898 endingPage "132" @default.
• W4308347898 startingPage "122" @default.
• W4308347898 abstract "Imperceptible adversarial examples are capable of deceiving the deep neural networks with high confidence. Recent studies show that it is particularly effective to control the attack space to low-frequency components of the image on the basis of adversarial training. However, those methods face the problem of losing valuable knowledge, especially shape information, which is vital for classification and robustness. To alleviate this issue, we propose a new method based on edge detection, named Edge Enhancement (EE), which can explicitly make up for the missing shape information in frequency constraints and further enhance the adversarial robustness. Specifically, we first employ a traditional edge detection algorithm called Canny to obtain shape information due to its simplicity and intrinsic robustness. Then, we augment the low-frequency space via obtained shape features, with the weighting operation carried on. This operation can be regarded as an emphasis on shape information, which could mitigate the texture bias of deep neural networks, thereby further serving the robustness. Finally, we feed the augmented features into the deep neural network. It is worth noting that these modules are optimized along with the deep neural network, which enables an end-to-end training fashion. Experimental results show that our proposed model can significantly improve adversarial robustness over the state-of-the-art methods on three benchmark datasets, including MNIST, Tiny ImageNet, and particularly ImageNet. For example, our method achieves 51.66% accuracy on ImageNet under 10-iteration targeted PGD white-box attack where the prior art has 36.94% accuracy. Code is available at https://github.com/Aiqz/Edge-Enhancement." @default.
• W4308347898 created "2022-11-11" @default.
• W4308347898 creator A5003763794 @default.
• W4308347898 creator A5004994916 @default.
• W4308347898 creator A5006340696 @default.
• W4308347898 creator A5051227924 @default.
• W4308347898 creator A5055826974 @default.
• W4308347898 creator A5085304489 @default.
• W4308347898 date "2023-01-01" @default.
• W4308347898 modified "2023-10-16" @default.
• W4308347898 title "Edge enhancement improves adversarial robustness in image classification" @default.
• W4308347898 cites W1980287119 @default.
• W4308347898 cites W1982374456 @default.
• W4308347898 cites W2009164871 @default.
• W4308347898 cites W2017745767 @default.
• W4308347898 cites W2031614119 @default.
• W4308347898 cites W2074355764 @default.
• W4308347898 cites W2117539524 @default.
• W4308347898 cites W2139399084 @default.
• W4308347898 cites W2145023731 @default.
• W4308347898 cites W2160815625 @default.
• W4308347898 cites W2183341477 @default.
• W4308347898 cites W2243397390 @default.
• W4308347898 cites W2290701356 @default.
• W4308347898 cites W2332769336 @default.
• W4308347898 cites W2530186022 @default.
• W4308347898 cites W2543927648 @default.
• W4308347898 cites W2586807479 @default.
• W4308347898 cites W2592929672 @default.
• W4308347898 cites W2603766943 @default.
• W4308347898 cites W2608070932 @default.
• W4308347898 cites W2746600820 @default.
• W4308347898 cites W2782360958 @default.
• W4308347898 cites W2804717358 @default.
• W4308347898 cites W2891656133 @default.
• W4308347898 cites W2912581782 @default.
• W4308347898 cites W2912697496 @default.
• W4308347898 cites W2923054028 @default.
• W4308347898 cites W2949066922 @default.
• W4308347898 cites W2952555815 @default.
• W4308347898 cites W2962872506 @default.
• W4308347898 cites W2987100326 @default.
• W4308347898 cites W3034537217 @default.
• W4308347898 cites W3034681682 @default.
• W4308347898 cites W3034894993 @default.
• W4308347898 cites W3120894151 @default.
• W4308347898 cites W4282918038 @default.
• W4308347898 doi "https://doi.org/10.1016/j.neucom.2022.10.059" @default.
• W4308347898 hasPublicationYear "2023" @default.
• W4308347898 type Work @default.
• W4308347898 citedByCount "1" @default.
• W4308347898 countsByYear W43083478982023 @default.
• W4308347898 crossrefType "journal-article" @default.
• W4308347898 hasAuthorship W4308347898A5003763794 @default.
• W4308347898 hasAuthorship W4308347898A5004994916 @default.
• W4308347898 hasAuthorship W4308347898A5006340696 @default.
• W4308347898 hasAuthorship W4308347898A5051227924 @default.
• W4308347898 hasAuthorship W4308347898A5055826974 @default.
• W4308347898 hasAuthorship W4308347898A5085304489 @default.
• W4308347898 hasConcept C104317684 @default.
• W4308347898 hasConcept C108583219 @default.
• W4308347898 hasConcept C115961682 @default.
• W4308347898 hasConcept C119857082 @default.
• W4308347898 hasConcept C126838900 @default.
• W4308347898 hasConcept C153180895 @default.
• W4308347898 hasConcept C154945302 @default.
• W4308347898 hasConcept C183115368 @default.
• W4308347898 hasConcept C185592680 @default.
• W4308347898 hasConcept C190502265 @default.
• W4308347898 hasConcept C22019652 @default.
• W4308347898 hasConcept C2984842247 @default.
• W4308347898 hasConcept C37736160 @default.
• W4308347898 hasConcept C41008148 @default.
• W4308347898 hasConcept C50644808 @default.
• W4308347898 hasConcept C55493867 @default.
• W4308347898 hasConcept C63479239 @default.
• W4308347898 hasConcept C71924100 @default.
• W4308347898 hasConcept C75294576 @default.
• W4308347898 hasConceptScore W4308347898C104317684 @default.
• W4308347898 hasConceptScore W4308347898C108583219 @default.
• W4308347898 hasConceptScore W4308347898C115961682 @default.
• W4308347898 hasConceptScore W4308347898C119857082 @default.
• W4308347898 hasConceptScore W4308347898C126838900 @default.
• W4308347898 hasConceptScore W4308347898C153180895 @default.
• W4308347898 hasConceptScore W4308347898C154945302 @default.
• W4308347898 hasConceptScore W4308347898C183115368 @default.
• W4308347898 hasConceptScore W4308347898C185592680 @default.
• W4308347898 hasConceptScore W4308347898C190502265 @default.
• W4308347898 hasConceptScore W4308347898C22019652 @default.
• W4308347898 hasConceptScore W4308347898C2984842247 @default.
• W4308347898 hasConceptScore W4308347898C37736160 @default.
• W4308347898 hasConceptScore W4308347898C41008148 @default.
• W4308347898 hasConceptScore W4308347898C50644808 @default.
• W4308347898 hasConceptScore W4308347898C55493867 @default.
• W4308347898 hasConceptScore W4308347898C63479239 @default.
• W4308347898 hasConceptScore W4308347898C71924100 @default.
• W4308347898 hasConceptScore W4308347898C75294576 @default.
• W4308347898 hasLocation W43083478981 @default.

• next