Matches in SemOpenAlex for { <https://semopenalex.org/work/W3208191520> ?p ?o ?g. }
Showing items 1 to 89 of
89
with 100 items per page.
- W3208191520 abstract "We study the problem of generating counterfactual text for a classifier as a means for understanding and debugging classification. Given a textual input and a classification model, we aim to minimally alter the text to change the model’s prediction. White-box approaches have been successfully applied to similar problems in vision where one can directly optimize the continuous input. Optimization-based approaches become difficult in the language domain due to the discrete nature of text. We bypass this issue by directly optimizing in the latent space and leveraging a language model to generate candidate modifications from optimized latent representations. We additionally use Shapley values to estimate the combinatoric effect of multiple changes. We then use these estimates to guide a beam search for the final counterfactual text. We achieve favorable performance compared to recent white-box and black-box baselines using human and automatic evaluations. Ablation studies show that both latent optimization and the use of Shapley values improve success rate and the quality of the generated counterfactuals." @default.
- W3208191520 created "2021-11-08" @default.
- W3208191520 creator A5064053803 @default.
- W3208191520 creator A5075188291 @default.
- W3208191520 date "2021-01-01" @default.
- W3208191520 modified "2023-09-30" @default.
- W3208191520 title "Text Counterfactuals via Latent Optimization and Shapley-Guided Search" @default.
- W3208191520 cites W2113459411 @default.
- W3208191520 cites W2121879602 @default.
- W3208191520 cites W2427527485 @default.
- W3208191520 cites W2618851150 @default.
- W3208191520 cites W2794557536 @default.
- W3208191520 cites W2799194071 @default.
- W3208191520 cites W2895752198 @default.
- W3208191520 cites W2896457183 @default.
- W3208191520 cites W2921189944 @default.
- W3208191520 cites W2949128310 @default.
- W3208191520 cites W2953958347 @default.
- W3208191520 cites W2963067999 @default.
- W3208191520 cites W2963207607 @default.
- W3208191520 cites W2963857521 @default.
- W3208191520 cites W2964222296 @default.
- W3208191520 cites W2964232431 @default.
- W3208191520 cites W2965373594 @default.
- W3208191520 cites W2970453125 @default.
- W3208191520 cites W2970562804 @default.
- W3208191520 cites W2970901330 @default.
- W3208191520 cites W2996851481 @default.
- W3208191520 cites W3014153848 @default.
- W3208191520 cites W3019166713 @default.
- W3208191520 cites W3034764457 @default.
- W3208191520 cites W3034873438 @default.
- W3208191520 cites W3035736465 @default.
- W3208191520 cites W3086385389 @default.
- W3208191520 cites W3092079550 @default.
- W3208191520 cites W3105604018 @default.
- W3208191520 cites W3114925695 @default.
- W3208191520 cites W3118796926 @default.
- W3208191520 doi "https://doi.org/10.18653/v1/2021.emnlp-main.452" @default.
- W3208191520 hasPublicationYear "2021" @default.
- W3208191520 type Work @default.
- W3208191520 sameAs 3208191520 @default.
- W3208191520 citedByCount "3" @default.
- W3208191520 countsByYear W32081915202022 @default.
- W3208191520 countsByYear W32081915202023 @default.
- W3208191520 crossrefType "proceedings-article" @default.
- W3208191520 hasAuthorship W3208191520A5064053803 @default.
- W3208191520 hasAuthorship W3208191520A5075188291 @default.
- W3208191520 hasBestOaLocation W32081915201 @default.
- W3208191520 hasConcept C108650721 @default.
- W3208191520 hasConcept C111472728 @default.
- W3208191520 hasConcept C119857082 @default.
- W3208191520 hasConcept C124101348 @default.
- W3208191520 hasConcept C138885662 @default.
- W3208191520 hasConcept C154945302 @default.
- W3208191520 hasConcept C168065819 @default.
- W3208191520 hasConcept C199360897 @default.
- W3208191520 hasConcept C41008148 @default.
- W3208191520 hasConcept C71889745 @default.
- W3208191520 hasConcept C95623464 @default.
- W3208191520 hasConceptScore W3208191520C108650721 @default.
- W3208191520 hasConceptScore W3208191520C111472728 @default.
- W3208191520 hasConceptScore W3208191520C119857082 @default.
- W3208191520 hasConceptScore W3208191520C124101348 @default.
- W3208191520 hasConceptScore W3208191520C138885662 @default.
- W3208191520 hasConceptScore W3208191520C154945302 @default.
- W3208191520 hasConceptScore W3208191520C168065819 @default.
- W3208191520 hasConceptScore W3208191520C199360897 @default.
- W3208191520 hasConceptScore W3208191520C41008148 @default.
- W3208191520 hasConceptScore W3208191520C71889745 @default.
- W3208191520 hasConceptScore W3208191520C95623464 @default.
- W3208191520 hasLocation W32081915201 @default.
- W3208191520 hasLocation W32081915202 @default.
- W3208191520 hasOpenAccess W3208191520 @default.
- W3208191520 hasPrimaryLocation W32081915201 @default.
- W3208191520 hasRelatedWork W2966062579 @default.
- W3208191520 hasRelatedWork W3089606255 @default.
- W3208191520 hasRelatedWork W4226198567 @default.
- W3208191520 hasRelatedWork W4282813524 @default.
- W3208191520 hasRelatedWork W4283721788 @default.
- W3208191520 hasRelatedWork W4285110382 @default.
- W3208191520 hasRelatedWork W4287889730 @default.
- W3208191520 hasRelatedWork W4288485742 @default.
- W3208191520 hasRelatedWork W4293223037 @default.
- W3208191520 hasRelatedWork W4310428066 @default.
- W3208191520 isParatext "false" @default.
- W3208191520 isRetracted "false" @default.
- W3208191520 magId "3208191520" @default.
- W3208191520 workType "article" @default.