Matches in SemOpenAlex for { <https://semopenalex.org/work/W3201816072> ?p ?o ?g. }
Showing items 1 to 89 of
89
with 100 items per page.
- W3201816072 abstract "Adversarial attacks/perturbations are becoming important for NLP research, as it has been shown recently that text-attacking adversaries can degrade an NLP model's performance without the victim's knowledge. This has far-reaching implications, especially when an NLP system is deployed in critical applications such as health or finance. In fact, the robustness of state-of-the-art models such as NLP transformers have increasingly been scrutinised due to their vulnerability against adversarial perturbations such as TextFooler and BERT-Attack. These methods, however, focus on changing words, which at times, ruins the readability and semantics of the sample. This paper introduces Special Symbol Text Attacks ‘SSTA’, a technique to improve the performance of language adversarial perturbations using special symbols that have downstream task information associated with them even though that should not have been the case. Our tests show that introducing such symbols which are meaningless to a human within a sentence, can perturb the sample in a particular direction. When this technique is used with TextFooler, which is a recent benchmark for the creation of NLP adversaries, through the TextAttack framework, it can improve all main evaluation metrics on three sentiment classification tasks and fake news detection. A simple, novel and symbol-specific adversarial learning technique is then introduced to reduce the influence of such special symbols." @default.
- W3201816072 created "2021-10-11" @default.
- W3201816072 creator A5001103949 @default.
- W3201816072 creator A5001319605 @default.
- W3201816072 creator A5090171111 @default.
- W3201816072 date "2021-07-18" @default.
- W3201816072 modified "2023-09-27" @default.
- W3201816072 title "Special Symbol Attacks On NLP Systems" @default.
- W3201816072 cites W2039117335 @default.
- W3201816072 cites W2163455955 @default.
- W3201816072 cites W2794557536 @default.
- W3201816072 cites W2798384383 @default.
- W3201816072 cites W2799194071 @default.
- W3201816072 cites W2897355816 @default.
- W3201816072 cites W2962718684 @default.
- W3201816072 cites W2963012544 @default.
- W3201816072 cites W2963207607 @default.
- W3201816072 cites W2963341956 @default.
- W3201816072 cites W2963661177 @default.
- W3201816072 cites W2965373594 @default.
- W3201816072 cites W2996851481 @default.
- W3201816072 cites W2997645422 @default.
- W3201816072 cites W3044324512 @default.
- W3201816072 cites W3092347010 @default.
- W3201816072 cites W3101449015 @default.
- W3201816072 cites W3103557498 @default.
- W3201816072 cites W3169948074 @default.
- W3201816072 doi "https://doi.org/10.1109/ijcnn52387.2021.9534254" @default.
- W3201816072 hasPublicationYear "2021" @default.
- W3201816072 type Work @default.
- W3201816072 sameAs 3201816072 @default.
- W3201816072 citedByCount "0" @default.
- W3201816072 crossrefType "proceedings-article" @default.
- W3201816072 hasAuthorship W3201816072A5001103949 @default.
- W3201816072 hasAuthorship W3201816072A5001319605 @default.
- W3201816072 hasAuthorship W3201816072A5090171111 @default.
- W3201816072 hasConcept C104317684 @default.
- W3201816072 hasConcept C119857082 @default.
- W3201816072 hasConcept C120665830 @default.
- W3201816072 hasConcept C121332964 @default.
- W3201816072 hasConcept C134400042 @default.
- W3201816072 hasConcept C137293760 @default.
- W3201816072 hasConcept C153083717 @default.
- W3201816072 hasConcept C154945302 @default.
- W3201816072 hasConcept C185592680 @default.
- W3201816072 hasConcept C192209626 @default.
- W3201816072 hasConcept C199360897 @default.
- W3201816072 hasConcept C204321447 @default.
- W3201816072 hasConcept C2777530160 @default.
- W3201816072 hasConcept C2778143727 @default.
- W3201816072 hasConcept C37736160 @default.
- W3201816072 hasConcept C41008148 @default.
- W3201816072 hasConcept C55493867 @default.
- W3201816072 hasConcept C63479239 @default.
- W3201816072 hasConceptScore W3201816072C104317684 @default.
- W3201816072 hasConceptScore W3201816072C119857082 @default.
- W3201816072 hasConceptScore W3201816072C120665830 @default.
- W3201816072 hasConceptScore W3201816072C121332964 @default.
- W3201816072 hasConceptScore W3201816072C134400042 @default.
- W3201816072 hasConceptScore W3201816072C137293760 @default.
- W3201816072 hasConceptScore W3201816072C153083717 @default.
- W3201816072 hasConceptScore W3201816072C154945302 @default.
- W3201816072 hasConceptScore W3201816072C185592680 @default.
- W3201816072 hasConceptScore W3201816072C192209626 @default.
- W3201816072 hasConceptScore W3201816072C199360897 @default.
- W3201816072 hasConceptScore W3201816072C204321447 @default.
- W3201816072 hasConceptScore W3201816072C2777530160 @default.
- W3201816072 hasConceptScore W3201816072C2778143727 @default.
- W3201816072 hasConceptScore W3201816072C37736160 @default.
- W3201816072 hasConceptScore W3201816072C41008148 @default.
- W3201816072 hasConceptScore W3201816072C55493867 @default.
- W3201816072 hasConceptScore W3201816072C63479239 @default.
- W3201816072 hasLocation W32018160721 @default.
- W3201816072 hasOpenAccess W3201816072 @default.
- W3201816072 hasPrimaryLocation W32018160721 @default.
- W3201816072 hasRelatedWork W1704713987 @default.
- W3201816072 hasRelatedWork W1978971213 @default.
- W3201816072 hasRelatedWork W2017590850 @default.
- W3201816072 hasRelatedWork W2169650257 @default.
- W3201816072 hasRelatedWork W2990196221 @default.
- W3201816072 hasRelatedWork W2997808760 @default.
- W3201816072 hasRelatedWork W3009120927 @default.
- W3201816072 hasRelatedWork W3107474891 @default.
- W3201816072 hasRelatedWork W4287890472 @default.
- W3201816072 hasRelatedWork W4311734044 @default.
- W3201816072 isParatext "false" @default.
- W3201816072 isRetracted "false" @default.
- W3201816072 magId "3201816072" @default.
- W3201816072 workType "article" @default.