Matches in SemOpenAlex for { <https://semopenalex.org/work/W4313120717> ?p ?o ?g. }
Showing items 1 to 92 of
92
with 100 items per page.
- W4313120717 abstract "Black-box adversarial attacks generate adversarial samples via iterative optimizations using repeated queries. Defending deep neural networks against such attacks has been challenging. In this paper, we propose an efficient Boundary Defense (BD) method which mitigates black-box attacks by exploiting the fact that the adversarial optimizations often need samples on the classification boundary. Our method detects the boundary samples as those with low classification confidence and adds white Gaussian noise to their logits. The method’s impact on the deep network’s classification accuracy is analyzed theoretically. Extensive experiments are conducted and the results show that the BD method can reliably defend against both soft and hard label black-box attacks. It outperforms a list of existing defense methods. For IMAGENET models, by adding zero-mean white Gaussian noise with standard deviation 0.1 to the logits of those images with classification confidence less than 0.3, the defense reduces the attack success rate to almost 0 while limiting the classification accuracy degradation to around 1 percent." @default.
- W4313120717 created "2023-01-06" @default.
- W4313120717 creator A5017805934 @default.
- W4313120717 creator A5071814939 @default.
- W4313120717 date "2022-08-21" @default.
- W4313120717 modified "2023-09-26" @default.
- W4313120717 title "Boundary Defense Against Black-box Adversarial Attacks" @default.
- W4313120717 cites W2160815625 @default.
- W4313120717 cites W2746600820 @default.
- W4313120717 cites W2754049786 @default.
- W4313120717 cites W2918288556 @default.
- W4313120717 cites W2962711307 @default.
- W4313120717 cites W2963485691 @default.
- W4313120717 cites W2963564844 @default.
- W4313120717 cites W2963693747 @default.
- W4313120717 cites W2963857521 @default.
- W4313120717 cites W2963952467 @default.
- W4313120717 cites W2964082701 @default.
- W4313120717 cites W2964205597 @default.
- W4313120717 cites W2966658324 @default.
- W4313120717 cites W3015625436 @default.
- W4313120717 cites W3080260826 @default.
- W4313120717 cites W3091857398 @default.
- W4313120717 cites W3106797537 @default.
- W4313120717 cites W3107235539 @default.
- W4313120717 cites W3122061855 @default.
- W4313120717 cites W3203257711 @default.
- W4313120717 cites W9657784 @default.
- W4313120717 doi "https://doi.org/10.1109/icpr56361.2022.9956476" @default.
- W4313120717 hasPublicationYear "2022" @default.
- W4313120717 type Work @default.
- W4313120717 citedByCount "1" @default.
- W4313120717 countsByYear W43131207172022 @default.
- W4313120717 crossrefType "proceedings-article" @default.
- W4313120717 hasAuthorship W4313120717A5017805934 @default.
- W4313120717 hasAuthorship W4313120717A5071814939 @default.
- W4313120717 hasBestOaLocation W43131207172 @default.
- W4313120717 hasConcept C11413529 @default.
- W4313120717 hasConcept C115961682 @default.
- W4313120717 hasConcept C119857082 @default.
- W4313120717 hasConcept C124101348 @default.
- W4313120717 hasConcept C127413603 @default.
- W4313120717 hasConcept C134306372 @default.
- W4313120717 hasConcept C153180895 @default.
- W4313120717 hasConcept C154945302 @default.
- W4313120717 hasConcept C188198153 @default.
- W4313120717 hasConcept C2984842247 @default.
- W4313120717 hasConcept C33923547 @default.
- W4313120717 hasConcept C37736160 @default.
- W4313120717 hasConcept C41008148 @default.
- W4313120717 hasConcept C4199805 @default.
- W4313120717 hasConcept C50644808 @default.
- W4313120717 hasConcept C62354387 @default.
- W4313120717 hasConcept C78519656 @default.
- W4313120717 hasConcept C94966114 @default.
- W4313120717 hasConcept C99498987 @default.
- W4313120717 hasConceptScore W4313120717C11413529 @default.
- W4313120717 hasConceptScore W4313120717C115961682 @default.
- W4313120717 hasConceptScore W4313120717C119857082 @default.
- W4313120717 hasConceptScore W4313120717C124101348 @default.
- W4313120717 hasConceptScore W4313120717C127413603 @default.
- W4313120717 hasConceptScore W4313120717C134306372 @default.
- W4313120717 hasConceptScore W4313120717C153180895 @default.
- W4313120717 hasConceptScore W4313120717C154945302 @default.
- W4313120717 hasConceptScore W4313120717C188198153 @default.
- W4313120717 hasConceptScore W4313120717C2984842247 @default.
- W4313120717 hasConceptScore W4313120717C33923547 @default.
- W4313120717 hasConceptScore W4313120717C37736160 @default.
- W4313120717 hasConceptScore W4313120717C41008148 @default.
- W4313120717 hasConceptScore W4313120717C4199805 @default.
- W4313120717 hasConceptScore W4313120717C50644808 @default.
- W4313120717 hasConceptScore W4313120717C62354387 @default.
- W4313120717 hasConceptScore W4313120717C78519656 @default.
- W4313120717 hasConceptScore W4313120717C94966114 @default.
- W4313120717 hasConceptScore W4313120717C99498987 @default.
- W4313120717 hasLocation W43131207171 @default.
- W4313120717 hasLocation W43131207172 @default.
- W4313120717 hasOpenAccess W4313120717 @default.
- W4313120717 hasPrimaryLocation W43131207171 @default.
- W4313120717 hasRelatedWork W2896078964 @default.
- W4313120717 hasRelatedWork W2914158293 @default.
- W4313120717 hasRelatedWork W2943646750 @default.
- W4313120717 hasRelatedWork W2963361074 @default.
- W4313120717 hasRelatedWork W3157170264 @default.
- W4313120717 hasRelatedWork W4210611492 @default.
- W4313120717 hasRelatedWork W4213432687 @default.
- W4313120717 hasRelatedWork W4221142553 @default.
- W4313120717 hasRelatedWork W4293054861 @default.
- W4313120717 hasRelatedWork W4313120717 @default.
- W4313120717 isParatext "false" @default.
- W4313120717 isRetracted "false" @default.
- W4313120717 workType "article" @default.