Matches in SemOpenAlex for { <https://semopenalex.org/work/W2901801467> ?p ?o ?g. }
- W2901801467 abstract "Researchers have observed that Visual Question Answering (VQA) models tend to answer questions by learning statistical biases in the data. For example, their answer to the question What is the color of the grass? is usually Green, whereas a question like What is the title of the book? cannot be answered by inferring statistical biases. It is of interest to the community to explicitly discover such biases, both for understanding the behavior of such models, and towards debugging them. Our work address this problem. In a database, we store the words of the question, answer and visual words corresponding to regions of interest in attention maps. By running simple rule mining algorithms on this database, we discover human-interpretable rules which give us unique insight into the behavior of such models. Our results also show examples of unusual behaviors learned by models in attempting VQA tasks." @default.
- W2901801467 created "2018-11-29" @default.
- W2901801467 creator A5026869260 @default.
- W2901801467 creator A5055264952 @default.
- W2901801467 creator A5066880080 @default.
- W2901801467 date "2018-11-19" @default.
- W2901801467 modified "2023-09-27" @default.
- W2901801467 title "Explicit Bias Discovery in Visual Question Answering Models" @default.
- W2901801467 cites W1506285740 @default.
- W2901801467 cites W1514535095 @default.
- W2901801467 cites W1624140875 @default.
- W2901801467 cites W1895577753 @default.
- W2901801467 cites W2064675550 @default.
- W2901801467 cites W2131846894 @default.
- W2901801467 cites W2194775991 @default.
- W2901801467 cites W2282821441 @default.
- W2901801467 cites W2412400526 @default.
- W2901801467 cites W2467818129 @default.
- W2901801467 cites W2560730294 @default.
- W2901801467 cites W2561196672 @default.
- W2901801467 cites W2561715562 @default.
- W2901801467 cites W2753300133 @default.
- W2901801467 cites W2763683025 @default.
- W2901801467 cites W2788481061 @default.
- W2901801467 cites W2789240707 @default.
- W2901801467 cites W2794583223 @default.
- W2901801467 cites W2798312813 @default.
- W2901801467 cites W2949474740 @default.
- W2901801467 cites W2950018712 @default.
- W2901801467 cites W2950538796 @default.
- W2901801467 cites W2950635152 @default.
- W2901801467 cites W2950761309 @default.
- W2901801467 cites W2951619830 @default.
- W2901801467 cites W2952122856 @default.
- W2901801467 cites W2952211726 @default.
- W2901801467 cites W2962787423 @default.
- W2901801467 cites W2962790223 @default.
- W2901801467 cites W2962858109 @default.
- W2901801467 cites W2963260436 @default.
- W2901801467 cites W2963644680 @default.
- W2901801467 cites W2963686699 @default.
- W2901801467 cites W2963717374 @default.
- W2901801467 cites W2963907708 @default.
- W2901801467 cites W2964118342 @default.
- W2901801467 hasPublicationYear "2018" @default.
- W2901801467 type Work @default.
- W2901801467 sameAs 2901801467 @default.
- W2901801467 citedByCount "6" @default.
- W2901801467 countsByYear W29018014672019 @default.
- W2901801467 countsByYear W29018014672020 @default.
- W2901801467 countsByYear W29018014672021 @default.
- W2901801467 crossrefType "posted-content" @default.
- W2901801467 hasAuthorship W2901801467A5026869260 @default.
- W2901801467 hasAuthorship W2901801467A5055264952 @default.
- W2901801467 hasAuthorship W2901801467A5066880080 @default.
- W2901801467 hasConcept C111472728 @default.
- W2901801467 hasConcept C119857082 @default.
- W2901801467 hasConcept C138885662 @default.
- W2901801467 hasConcept C154945302 @default.
- W2901801467 hasConcept C168065819 @default.
- W2901801467 hasConcept C199360897 @default.
- W2901801467 hasConcept C204321447 @default.
- W2901801467 hasConcept C23123220 @default.
- W2901801467 hasConcept C2522767166 @default.
- W2901801467 hasConcept C2780586882 @default.
- W2901801467 hasConcept C41008148 @default.
- W2901801467 hasConcept C44291984 @default.
- W2901801467 hasConceptScore W2901801467C111472728 @default.
- W2901801467 hasConceptScore W2901801467C119857082 @default.
- W2901801467 hasConceptScore W2901801467C138885662 @default.
- W2901801467 hasConceptScore W2901801467C154945302 @default.
- W2901801467 hasConceptScore W2901801467C168065819 @default.
- W2901801467 hasConceptScore W2901801467C199360897 @default.
- W2901801467 hasConceptScore W2901801467C204321447 @default.
- W2901801467 hasConceptScore W2901801467C23123220 @default.
- W2901801467 hasConceptScore W2901801467C2522767166 @default.
- W2901801467 hasConceptScore W2901801467C2780586882 @default.
- W2901801467 hasConceptScore W2901801467C41008148 @default.
- W2901801467 hasConceptScore W2901801467C44291984 @default.
- W2901801467 hasLocation W29018014671 @default.
- W2901801467 hasOpenAccess W2901801467 @default.
- W2901801467 hasPrimaryLocation W29018014671 @default.
- W2901801467 hasRelatedWork W2252077675 @default.
- W2901801467 hasRelatedWork W2273038706 @default.
- W2901801467 hasRelatedWork W2391839782 @default.
- W2901801467 hasRelatedWork W2438044634 @default.
- W2901801467 hasRelatedWork W2521737809 @default.
- W2901801467 hasRelatedWork W2787553714 @default.
- W2901801467 hasRelatedWork W2790248406 @default.
- W2901801467 hasRelatedWork W2802581042 @default.
- W2901801467 hasRelatedWork W2900693018 @default.
- W2901801467 hasRelatedWork W2963456402 @default.
- W2901801467 hasRelatedWork W2982699810 @default.
- W2901801467 hasRelatedWork W3037215679 @default.
- W2901801467 hasRelatedWork W3045060014 @default.
- W2901801467 hasRelatedWork W3103706239 @default.
- W2901801467 hasRelatedWork W3113204237 @default.
- W2901801467 hasRelatedWork W3121918918 @default.
- W2901801467 hasRelatedWork W3122708867 @default.
- W2901801467 hasRelatedWork W3157008829 @default.