Matches in SemOpenAlex for { <https://semopenalex.org/work/W4385569501> ?p ?o ?g. }
Showing items 1 to 93 of
93
with 100 items per page.
- W4385569501 endingPage "3151" @default.
- W4385569501 startingPage "3139" @default.
- W4385569501 abstract "In text classification, the traditional attention mechanisms usually focus too much on frequent words, and need extensive labeled data in order to learn. This paper proposes a perturbation-based self-supervised attention approach to guide attention learning without any annotation overhead. Specifically, we add as much noise as possible to all the words in the sentence without changing their semantics and predictions. We hypothesize that words that tolerate more noise are less significant, and we can use this information to refine the attention distribution. Experimental results on three text classification tasks show that our approach can significantly improve the performance of current attention-based models, and is more effective than existing self-supervised methods. We also provide a visualization analysis to verify the effectiveness of our approach." @default.
- W4385569501 created "2023-08-05" @default.
- W4385569501 creator A5076609933 @default.
- W4385569501 creator A5079618579 @default.
- W4385569501 creator A5080125914 @default.
- W4385569501 date "2023-01-01" @default.
- W4385569501 modified "2023-10-17" @default.
- W4385569501 title "Perturbation-Based Self-Supervised Attention for Attention Bias in Text Classification" @default.
- W4385569501 cites W1902237438 @default.
- W4385569501 cites W2014902591 @default.
- W4385569501 cites W2070246124 @default.
- W4385569501 cites W2114524997 @default.
- W4385569501 cites W2160660844 @default.
- W4385569501 cites W2163455955 @default.
- W4385569501 cites W2251124635 @default.
- W4385569501 cites W2251648804 @default.
- W4385569501 cites W2470673105 @default.
- W4385569501 cites W2517194566 @default.
- W4385569501 cites W2608787653 @default.
- W4385569501 cites W2799044502 @default.
- W4385569501 cites W2869036160 @default.
- W4385569501 cites W2889436406 @default.
- W4385569501 cites W2898936689 @default.
- W4385569501 cites W2949161734 @default.
- W4385569501 cites W2950768109 @default.
- W4385569501 cites W2963082289 @default.
- W4385569501 cites W2963168371 @default.
- W4385569501 cites W2963240575 @default.
- W4385569501 cites W2963561255 @default.
- W4385569501 cites W3003410553 @default.
- W4385569501 cites W3034444624 @default.
- W4385569501 cites W3034917890 @default.
- W4385569501 cites W3101287165 @default.
- W4385569501 cites W3103099263 @default.
- W4385569501 cites W3104564752 @default.
- W4385569501 cites W3106388706 @default.
- W4385569501 cites W3116500963 @default.
- W4385569501 cites W3130441769 @default.
- W4385569501 cites W3167369375 @default.
- W4385569501 cites W3170572542 @default.
- W4385569501 doi "https://doi.org/10.1109/taslp.2023.3302230" @default.
- W4385569501 hasPublicationYear "2023" @default.
- W4385569501 type Work @default.
- W4385569501 citedByCount "0" @default.
- W4385569501 crossrefType "journal-article" @default.
- W4385569501 hasAuthorship W4385569501A5076609933 @default.
- W4385569501 hasAuthorship W4385569501A5079618579 @default.
- W4385569501 hasAuthorship W4385569501A5080125914 @default.
- W4385569501 hasBestOaLocation W43855695012 @default.
- W4385569501 hasConcept C119857082 @default.
- W4385569501 hasConcept C136389625 @default.
- W4385569501 hasConcept C154945302 @default.
- W4385569501 hasConcept C184337299 @default.
- W4385569501 hasConcept C199360897 @default.
- W4385569501 hasConcept C204321447 @default.
- W4385569501 hasConcept C2776321320 @default.
- W4385569501 hasConcept C2777530160 @default.
- W4385569501 hasConcept C36464697 @default.
- W4385569501 hasConcept C41008148 @default.
- W4385569501 hasConcept C50644808 @default.
- W4385569501 hasConceptScore W4385569501C119857082 @default.
- W4385569501 hasConceptScore W4385569501C136389625 @default.
- W4385569501 hasConceptScore W4385569501C154945302 @default.
- W4385569501 hasConceptScore W4385569501C184337299 @default.
- W4385569501 hasConceptScore W4385569501C199360897 @default.
- W4385569501 hasConceptScore W4385569501C204321447 @default.
- W4385569501 hasConceptScore W4385569501C2776321320 @default.
- W4385569501 hasConceptScore W4385569501C2777530160 @default.
- W4385569501 hasConceptScore W4385569501C36464697 @default.
- W4385569501 hasConceptScore W4385569501C41008148 @default.
- W4385569501 hasConceptScore W4385569501C50644808 @default.
- W4385569501 hasFunder F4320321001 @default.
- W4385569501 hasFunder F4320321921 @default.
- W4385569501 hasLocation W43855695011 @default.
- W4385569501 hasLocation W43855695012 @default.
- W4385569501 hasOpenAccess W4385569501 @default.
- W4385569501 hasPrimaryLocation W43855695011 @default.
- W4385569501 hasRelatedWork W1533177136 @default.
- W4385569501 hasRelatedWork W2218034408 @default.
- W4385569501 hasRelatedWork W2263699433 @default.
- W4385569501 hasRelatedWork W2358755282 @default.
- W4385569501 hasRelatedWork W2361861616 @default.
- W4385569501 hasRelatedWork W2377979023 @default.
- W4385569501 hasRelatedWork W2392921965 @default.
- W4385569501 hasRelatedWork W2556260348 @default.
- W4385569501 hasRelatedWork W2625833328 @default.
- W4385569501 hasRelatedWork W4380994516 @default.
- W4385569501 hasVolume "31" @default.
- W4385569501 isParatext "false" @default.
- W4385569501 isRetracted "false" @default.
- W4385569501 workType "article" @default.