Matches in SemOpenAlex for { <https://semopenalex.org/work/W3015909098> ?p ?o ?g. }
Showing items 1 to 86 of
86
with 100 items per page.
- W3015909098 abstract "Recent advances in large-scale language representation models such as BERT have improved the state-of-the-art performances in many NLP tasks. Meanwhile, character-level Chinese NLP models, including BERT for Chinese, have also demonstrated that they can outperform the existing models. In this paper, we show that, however, such BERT-based models are vulnerable under character-level adversarial attacks. We propose a novel Chinese char-level attack method against BERT-based classifiers. Essentially, we generate small perturbation on the character level in the embedding space and guide the character substitution procedure. Extensive experiments show that the classification accuracy on a Chinese news dataset drops from 91.8% to 0% by manipulating less than 2 characters on average based on the proposed attack. Human evaluations also confirm that our generated Chinese adversarial examples barely affect human performance on these NLP tasks." @default.
- W3015909098 created "2020-04-17" @default.
- W3015909098 creator A5029029467 @default.
- W3015909098 creator A5041191241 @default.
- W3015909098 creator A5079034292 @default.
- W3015909098 creator A5090592237 @default.
- W3015909098 date "2020-04-07" @default.
- W3015909098 modified "2023-09-26" @default.
- W3015909098 title "Towards Evaluating the Robustness of Chinese BERT Classifiers" @default.
- W3015909098 cites W2626778328 @default.
- W3015909098 cites W2890042874 @default.
- W3015909098 cites W2892852825 @default.
- W3015909098 cites W2896457183 @default.
- W3015909098 cites W2922266396 @default.
- W3015909098 cites W2944852028 @default.
- W3015909098 cites W2949128310 @default.
- W3015909098 cites W2950606982 @default.
- W3015909098 cites W2952595356 @default.
- W3015909098 cites W2952939310 @default.
- W3015909098 cites W2953307569 @default.
- W3015909098 cites W2953356739 @default.
- W3015909098 cites W2962718684 @default.
- W3015909098 cites W2963857521 @default.
- W3015909098 cites W2963859254 @default.
- W3015909098 cites W2963969878 @default.
- W3015909098 cites W2964121744 @default.
- W3015909098 cites W2964301649 @default.
- W3015909098 cites W2970597249 @default.
- W3015909098 cites W2975059944 @default.
- W3015909098 doi "https://doi.org/10.48550/arxiv.2004.03742" @default.
- W3015909098 hasPublicationYear "2020" @default.
- W3015909098 type Work @default.
- W3015909098 sameAs 3015909098 @default.
- W3015909098 citedByCount "1" @default.
- W3015909098 countsByYear W30159090982019 @default.
- W3015909098 crossrefType "posted-content" @default.
- W3015909098 hasAuthorship W3015909098A5029029467 @default.
- W3015909098 hasAuthorship W3015909098A5041191241 @default.
- W3015909098 hasAuthorship W3015909098A5079034292 @default.
- W3015909098 hasAuthorship W3015909098A5090592237 @default.
- W3015909098 hasBestOaLocation W30159090981 @default.
- W3015909098 hasConcept C104317684 @default.
- W3015909098 hasConcept C119857082 @default.
- W3015909098 hasConcept C137293760 @default.
- W3015909098 hasConcept C154945302 @default.
- W3015909098 hasConcept C185592680 @default.
- W3015909098 hasConcept C204321447 @default.
- W3015909098 hasConcept C2524010 @default.
- W3015909098 hasConcept C2780861071 @default.
- W3015909098 hasConcept C33923547 @default.
- W3015909098 hasConcept C37736160 @default.
- W3015909098 hasConcept C41008148 @default.
- W3015909098 hasConcept C41608201 @default.
- W3015909098 hasConcept C55493867 @default.
- W3015909098 hasConcept C63479239 @default.
- W3015909098 hasConceptScore W3015909098C104317684 @default.
- W3015909098 hasConceptScore W3015909098C119857082 @default.
- W3015909098 hasConceptScore W3015909098C137293760 @default.
- W3015909098 hasConceptScore W3015909098C154945302 @default.
- W3015909098 hasConceptScore W3015909098C185592680 @default.
- W3015909098 hasConceptScore W3015909098C204321447 @default.
- W3015909098 hasConceptScore W3015909098C2524010 @default.
- W3015909098 hasConceptScore W3015909098C2780861071 @default.
- W3015909098 hasConceptScore W3015909098C33923547 @default.
- W3015909098 hasConceptScore W3015909098C37736160 @default.
- W3015909098 hasConceptScore W3015909098C41008148 @default.
- W3015909098 hasConceptScore W3015909098C41608201 @default.
- W3015909098 hasConceptScore W3015909098C55493867 @default.
- W3015909098 hasConceptScore W3015909098C63479239 @default.
- W3015909098 hasLocation W30159090981 @default.
- W3015909098 hasOpenAccess W3015909098 @default.
- W3015909098 hasPrimaryLocation W30159090981 @default.
- W3015909098 hasRelatedWork W2952919291 @default.
- W3015909098 hasRelatedWork W2978686724 @default.
- W3015909098 hasRelatedWork W3035729345 @default.
- W3015909098 hasRelatedWork W3107474891 @default.
- W3015909098 hasRelatedWork W3193857078 @default.
- W3015909098 hasRelatedWork W3208304128 @default.
- W3015909098 hasRelatedWork W4287549955 @default.
- W3015909098 hasRelatedWork W4306177400 @default.
- W3015909098 hasRelatedWork W4311734044 @default.
- W3015909098 hasRelatedWork W4372348577 @default.
- W3015909098 isParatext "false" @default.
- W3015909098 isRetracted "false" @default.
- W3015909098 magId "3015909098" @default.
- W3015909098 workType "article" @default.