Matches in SemOpenAlex for { <https://semopenalex.org/work/W3217604114> ?p ?o ?g. }
- W3217604114 endingPage "111" @default.
- W3217604114 startingPage "98" @default.
- W3217604114 abstract "Over the past few years, various word-level textual attack approaches have been proposed to reveal the vulnerability of deep neural networks used in natural language processing. Typically, these approaches involve an important optimization step to determine which substitute to be used for each word in the original input. However, current research on this step is still rather limited, from the perspectives of both problem-understanding and problem-solving. In this paper, we address these issues by uncovering the theoretical properties of the problem and proposing an efficient local search algorithm (LS) to solve it. We establish the <italic xmlns:mml=http://www.w3.org/1998/Math/MathML xmlns:xlink=http://www.w3.org/1999/xlink>first</i> provable approximation guarantee on solving the problem in general cases. Extensive experiments involving 5 NLP tasks, 8 datasets and 26 NLP models show that LS can largely reduce the number of queries usually by an order of magnitude to achieve high attack success rates. Further experiments show that the adversarial examples crafted by LS usually have higher quality, exhibit better transferability, and can bring more robustness improvement to victim models by adversarial training." @default.
- W3217604114 created "2021-12-06" @default.
- W3217604114 creator A5011778476 @default.
- W3217604114 creator A5017933282 @default.
- W3217604114 creator A5021254405 @default.
- W3217604114 creator A5047453825 @default.
- W3217604114 date "2022-01-01" @default.
- W3217604114 modified "2023-09-23" @default.
- W3217604114 title "Efficient Combinatorial Optimization for Word-Level Adversarial Textual Attack" @default.
- W3217604114 cites W1680189815 @default.
- W3217604114 cites W1936538920 @default.
- W3217604114 cites W1973315504 @default.
- W3217604114 cites W2599823825 @default.
- W3217604114 cites W2609368435 @default.
- W3217604114 cites W2766554185 @default.
- W3217604114 cites W2774644650 @default.
- W3217604114 cites W2799007037 @default.
- W3217604114 cites W2799194071 @default.
- W3217604114 cites W2947415936 @default.
- W3217604114 cites W2949128310 @default.
- W3217604114 cites W2962718684 @default.
- W3217604114 cites W2962818281 @default.
- W3217604114 cites W2963083752 @default.
- W3217604114 cites W2963126845 @default.
- W3217604114 cites W2963177632 @default.
- W3217604114 cites W2963481228 @default.
- W3217604114 cites W2963834268 @default.
- W3217604114 cites W2963859254 @default.
- W3217604114 cites W2963969878 @default.
- W3217604114 cites W2982756474 @default.
- W3217604114 cites W2996851481 @default.
- W3217604114 cites W2997524296 @default.
- W3217604114 cites W2998277219 @default.
- W3217604114 cites W3021907668 @default.
- W3217604114 cites W3023553115 @default.
- W3217604114 cites W3035736465 @default.
- W3217604114 cites W3101449015 @default.
- W3217604114 cites W3105604018 @default.
- W3217604114 cites W3132404260 @default.
- W3217604114 cites W4288953700 @default.
- W3217604114 doi "https://doi.org/10.1109/taslp.2021.3130970" @default.
- W3217604114 hasPublicationYear "2022" @default.
- W3217604114 type Work @default.
- W3217604114 sameAs 3217604114 @default.
- W3217604114 citedByCount "6" @default.
- W3217604114 countsByYear W32176041142022 @default.
- W3217604114 countsByYear W32176041142023 @default.
- W3217604114 crossrefType "journal-article" @default.
- W3217604114 hasAuthorship W3217604114A5011778476 @default.
- W3217604114 hasAuthorship W3217604114A5017933282 @default.
- W3217604114 hasAuthorship W3217604114A5021254405 @default.
- W3217604114 hasAuthorship W3217604114A5047453825 @default.
- W3217604114 hasBestOaLocation W32176041142 @default.
- W3217604114 hasConcept C104317684 @default.
- W3217604114 hasConcept C119857082 @default.
- W3217604114 hasConcept C140331021 @default.
- W3217604114 hasConcept C154945302 @default.
- W3217604114 hasConcept C185592680 @default.
- W3217604114 hasConcept C204321447 @default.
- W3217604114 hasConcept C2524010 @default.
- W3217604114 hasConcept C33923547 @default.
- W3217604114 hasConcept C37736160 @default.
- W3217604114 hasConcept C41008148 @default.
- W3217604114 hasConcept C55493867 @default.
- W3217604114 hasConcept C61272859 @default.
- W3217604114 hasConcept C63479239 @default.
- W3217604114 hasConcept C80444323 @default.
- W3217604114 hasConcept C90805587 @default.
- W3217604114 hasConceptScore W3217604114C104317684 @default.
- W3217604114 hasConceptScore W3217604114C119857082 @default.
- W3217604114 hasConceptScore W3217604114C140331021 @default.
- W3217604114 hasConceptScore W3217604114C154945302 @default.
- W3217604114 hasConceptScore W3217604114C185592680 @default.
- W3217604114 hasConceptScore W3217604114C204321447 @default.
- W3217604114 hasConceptScore W3217604114C2524010 @default.
- W3217604114 hasConceptScore W3217604114C33923547 @default.
- W3217604114 hasConceptScore W3217604114C37736160 @default.
- W3217604114 hasConceptScore W3217604114C41008148 @default.
- W3217604114 hasConceptScore W3217604114C55493867 @default.
- W3217604114 hasConceptScore W3217604114C61272859 @default.
- W3217604114 hasConceptScore W3217604114C63479239 @default.
- W3217604114 hasConceptScore W3217604114C80444323 @default.
- W3217604114 hasConceptScore W3217604114C90805587 @default.
- W3217604114 hasFunder F4320321885 @default.
- W3217604114 hasLocation W32176041141 @default.
- W3217604114 hasLocation W32176041142 @default.
- W3217604114 hasOpenAccess W3217604114 @default.
- W3217604114 hasPrimaryLocation W32176041141 @default.
- W3217604114 hasRelatedWork W2911510572 @default.
- W3217604114 hasRelatedWork W2938867157 @default.
- W3217604114 hasRelatedWork W3025658176 @default.
- W3217604114 hasRelatedWork W3087721806 @default.
- W3217604114 hasRelatedWork W3176127176 @default.
- W3217604114 hasRelatedWork W3185166384 @default.
- W3217604114 hasRelatedWork W4224329172 @default.
- W3217604114 hasRelatedWork W4226402597 @default.
- W3217604114 hasRelatedWork W4287663567 @default.
- W3217604114 hasRelatedWork W4288373550 @default.