Matches in SemOpenAlex for { <https://semopenalex.org/work/W2983256121> ?p ?o ?g. }
Showing items 1 to 96 of
96
with 100 items per page.
- W2983256121 abstract "Many vision and language models suffer from poor visual grounding -- often falling back on easy-to-learn language priors rather than basing their decisions on visual concepts in the image. In this work, we propose a generic approach called Human Importance-aware Network Tuning (HINT) that effectively leverages human demonstrations to improve visual grounding. HINT encourages deep networks to be sensitive to the same input regions as humans. Our approach optimizes the alignment between human attention maps and gradient-based network importances -- ensuring that models learn not just to look at but rather rely on visual concepts that humans found relevant for a task when making predictions. We apply HINT to Visual Question Answering and Image Captioning tasks, outperforming top approaches on splits that penalize over-reliance on language priors (VQA-CP and robust captioning) using human attention demonstrations for just 6% of the training data." @default.
- W2983256121 created "2019-11-22" @default.
- W2983256121 creator A5003679010 @default.
- W2983256121 creator A5014035752 @default.
- W2983256121 creator A5024902016 @default.
- W2983256121 creator A5044205212 @default.
- W2983256121 creator A5046238088 @default.
- W2983256121 creator A5050342343 @default.
- W2983256121 creator A5051259505 @default.
- W2983256121 creator A5058974318 @default.
- W2983256121 date "2019-10-01" @default.
- W2983256121 modified "2023-10-06" @default.
- W2983256121 title "Taking a HINT: Leveraging Explanations to Make Vision and Language Models More Grounded" @default.
- W2983256121 cites W1536680647 @default.
- W2983256121 cites W1931639407 @default.
- W2983256121 cites W1933349210 @default.
- W2983256121 cites W2157331557 @default.
- W2983256121 cites W2302086703 @default.
- W2983256121 cites W2560730294 @default.
- W2983256121 cites W2561715562 @default.
- W2983256121 cites W2575842049 @default.
- W2983256121 cites W2745461083 @default.
- W2983256121 cites W2759653627 @default.
- W2983256121 cites W2795151422 @default.
- W2983256121 cites W2886201417 @default.
- W2983256121 cites W2962858109 @default.
- W2983256121 cites W2963191264 @default.
- W2983256121 cites W2963260436 @default.
- W2983256121 cites W2963890019 @default.
- W2983256121 cites W2963954913 @default.
- W2983256121 cites W2983256121 @default.
- W2983256121 cites W3037725825 @default.
- W2983256121 doi "https://doi.org/10.1109/iccv.2019.00268" @default.
- W2983256121 hasPublicationYear "2019" @default.
- W2983256121 type Work @default.
- W2983256121 sameAs 2983256121 @default.
- W2983256121 citedByCount "150" @default.
- W2983256121 countsByYear W29832561212019 @default.
- W2983256121 countsByYear W29832561212020 @default.
- W2983256121 countsByYear W29832561212021 @default.
- W2983256121 countsByYear W29832561212022 @default.
- W2983256121 countsByYear W29832561212023 @default.
- W2983256121 crossrefType "proceedings-article" @default.
- W2983256121 hasAuthorship W2983256121A5003679010 @default.
- W2983256121 hasAuthorship W2983256121A5014035752 @default.
- W2983256121 hasAuthorship W2983256121A5024902016 @default.
- W2983256121 hasAuthorship W2983256121A5044205212 @default.
- W2983256121 hasAuthorship W2983256121A5046238088 @default.
- W2983256121 hasAuthorship W2983256121A5050342343 @default.
- W2983256121 hasAuthorship W2983256121A5051259505 @default.
- W2983256121 hasAuthorship W2983256121A5058974318 @default.
- W2983256121 hasBestOaLocation W29832561212 @default.
- W2983256121 hasConcept C107673813 @default.
- W2983256121 hasConcept C115961682 @default.
- W2983256121 hasConcept C119857082 @default.
- W2983256121 hasConcept C137293760 @default.
- W2983256121 hasConcept C154945302 @default.
- W2983256121 hasConcept C157657479 @default.
- W2983256121 hasConcept C162324750 @default.
- W2983256121 hasConcept C177769412 @default.
- W2983256121 hasConcept C187736073 @default.
- W2983256121 hasConcept C204321447 @default.
- W2983256121 hasConcept C2780451532 @default.
- W2983256121 hasConcept C41008148 @default.
- W2983256121 hasConcept C44291984 @default.
- W2983256121 hasConceptScore W2983256121C107673813 @default.
- W2983256121 hasConceptScore W2983256121C115961682 @default.
- W2983256121 hasConceptScore W2983256121C119857082 @default.
- W2983256121 hasConceptScore W2983256121C137293760 @default.
- W2983256121 hasConceptScore W2983256121C154945302 @default.
- W2983256121 hasConceptScore W2983256121C157657479 @default.
- W2983256121 hasConceptScore W2983256121C162324750 @default.
- W2983256121 hasConceptScore W2983256121C177769412 @default.
- W2983256121 hasConceptScore W2983256121C187736073 @default.
- W2983256121 hasConceptScore W2983256121C204321447 @default.
- W2983256121 hasConceptScore W2983256121C2780451532 @default.
- W2983256121 hasConceptScore W2983256121C41008148 @default.
- W2983256121 hasConceptScore W2983256121C44291984 @default.
- W2983256121 hasLocation W29832561211 @default.
- W2983256121 hasLocation W29832561212 @default.
- W2983256121 hasOpenAccess W2983256121 @default.
- W2983256121 hasPrimaryLocation W29832561211 @default.
- W2983256121 hasRelatedWork W1938708284 @default.
- W2983256121 hasRelatedWork W2964413124 @default.
- W2983256121 hasRelatedWork W3113264705 @default.
- W2983256121 hasRelatedWork W3122720459 @default.
- W2983256121 hasRelatedWork W3164229987 @default.
- W2983256121 hasRelatedWork W3204607391 @default.
- W2983256121 hasRelatedWork W3215212336 @default.
- W2983256121 hasRelatedWork W4290852288 @default.
- W2983256121 hasRelatedWork W4298897568 @default.
- W2983256121 hasRelatedWork W4380190185 @default.
- W2983256121 isParatext "false" @default.
- W2983256121 isRetracted "false" @default.
- W2983256121 magId "2983256121" @default.
- W2983256121 workType "article" @default.