FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W2903036216> ?p ?o ?g. }

• W2903036216 abstract "RNN-based sequence generation is now widely used in NLP and NLG (natural language generation). Most work focusses on how to train RNNs, even though also decoding is not necessarily straightforward: previous work on neural MT found seq2seq models to radically prefer short candidates, and has proposed a number of beam search heuristics to deal with this. In this work, we assess decoding strategies for referring expression generation with neural models. Here, expression length is crucial: output should neither contain too much or too little information, in order to be pragmatically adequate. We find that most beam search heuristics developed for MT do not generalize well to referring expression generation (REG), and do not generally outperform greedy decoding. We observe that beam search heuristics for termination seem to override the model’s knowledge of what a good stopping point is. Therefore, we also explore a recent approach called trainable decoding, which uses a small network to modify the RNN’s hidden state for better decoding results. We find this approach to consistently outperform greedy decoding for REG." @default.
• W2903036216 created "2018-12-11" @default.
• W2903036216 creator A5032801642 @default.
• W2903036216 creator A5078051602 @default.
• W2903036216 date "2018-01-01" @default.
• W2903036216 modified "2023-09-24" @default.
• W2903036216 title "Decoding Strategies for Neural Referring Expression Generation" @default.
• W2903036216 cites W1514535095 @default.
• W2903036216 cites W15689187 @default.
• W2903036216 cites W1578823404 @default.
• W2903036216 cites W1591706642 @default.
• W2903036216 cites W1645937837 @default.
• W2903036216 cites W1686810756 @default.
• W2903036216 cites W1828163288 @default.
• W2903036216 cites W1861492603 @default.
• W2903036216 cites W1889081078 @default.
• W2903036216 cites W1956340063 @default.
• W2903036216 cites W1983707534 @default.
• W2903036216 cites W1993800891 @default.
• W2903036216 cites W2072328995 @default.
• W2903036216 cites W2096646663 @default.
• W2903036216 cites W2099978801 @default.
• W2903036216 cites W2101105183 @default.
• W2903036216 cites W2125447031 @default.
• W2903036216 cites W2250351336 @default.
• W2903036216 cites W2251512949 @default.
• W2903036216 cites W2296073425 @default.
• W2903036216 cites W2489434015 @default.
• W2903036216 cites W2508818207 @default.
• W2903036216 cites W2525778437 @default.
• W2903036216 cites W2557436004 @default.
• W2903036216 cites W2571175805 @default.
• W2903036216 cites W2574790321 @default.
• W2903036216 cites W2757222607 @default.
• W2903036216 cites W2901909302 @default.
• W2903036216 cites W2963109634 @default.
• W2903036216 cites W2963141266 @default.
• W2903036216 cites W2963212250 @default.
• W2903036216 cites W2963248296 @default.
• W2903036216 cites W2964195418 @default.
• W2903036216 cites W2964308564 @default.
• W2903036216 cites W3102516861 @default.
• W2903036216 cites W2250206255 @default.
• W2903036216 doi "https://doi.org/10.18653/v1/w18-6563" @default.
• W2903036216 hasPublicationYear "2018" @default.
• W2903036216 type Work @default.
• W2903036216 sameAs 2903036216 @default.
• W2903036216 citedByCount "11" @default.
• W2903036216 countsByYear W29030362162019 @default.
• W2903036216 countsByYear W29030362162020 @default.
• W2903036216 countsByYear W29030362162021 @default.
• W2903036216 countsByYear W29030362162023 @default.
• W2903036216 crossrefType "proceedings-article" @default.
• W2903036216 hasAuthorship W2903036216A5032801642 @default.
• W2903036216 hasAuthorship W2903036216A5078051602 @default.
• W2903036216 hasBestOaLocation W29030362161 @default.
• W2903036216 hasConcept C111919701 @default.
• W2903036216 hasConcept C11413529 @default.
• W2903036216 hasConcept C125583679 @default.
• W2903036216 hasConcept C127705205 @default.
• W2903036216 hasConcept C137293760 @default.
• W2903036216 hasConcept C147168706 @default.
• W2903036216 hasConcept C154945302 @default.
• W2903036216 hasConcept C157125643 @default.
• W2903036216 hasConcept C195324797 @default.
• W2903036216 hasConcept C19889080 @default.
• W2903036216 hasConcept C199360897 @default.
• W2903036216 hasConcept C204397858 @default.
• W2903036216 hasConcept C2776187449 @default.
• W2903036216 hasConcept C28490314 @default.
• W2903036216 hasConcept C40743351 @default.
• W2903036216 hasConcept C41008148 @default.
• W2903036216 hasConcept C50644808 @default.
• W2903036216 hasConcept C51823790 @default.
• W2903036216 hasConcept C57273362 @default.
• W2903036216 hasConcept C78944582 @default.
• W2903036216 hasConcept C90559484 @default.
• W2903036216 hasConceptScore W2903036216C111919701 @default.
• W2903036216 hasConceptScore W2903036216C11413529 @default.
• W2903036216 hasConceptScore W2903036216C125583679 @default.
• W2903036216 hasConceptScore W2903036216C127705205 @default.
• W2903036216 hasConceptScore W2903036216C137293760 @default.
• W2903036216 hasConceptScore W2903036216C147168706 @default.
• W2903036216 hasConceptScore W2903036216C154945302 @default.
• W2903036216 hasConceptScore W2903036216C157125643 @default.
• W2903036216 hasConceptScore W2903036216C195324797 @default.
• W2903036216 hasConceptScore W2903036216C19889080 @default.
• W2903036216 hasConceptScore W2903036216C199360897 @default.
• W2903036216 hasConceptScore W2903036216C204397858 @default.
• W2903036216 hasConceptScore W2903036216C2776187449 @default.
• W2903036216 hasConceptScore W2903036216C28490314 @default.
• W2903036216 hasConceptScore W2903036216C40743351 @default.
• W2903036216 hasConceptScore W2903036216C41008148 @default.
• W2903036216 hasConceptScore W2903036216C50644808 @default.
• W2903036216 hasConceptScore W2903036216C51823790 @default.
• W2903036216 hasConceptScore W2903036216C57273362 @default.
• W2903036216 hasConceptScore W2903036216C78944582 @default.
• W2903036216 hasConceptScore W2903036216C90559484 @default.
• W2903036216 hasLocation W29030362161 @default.
• W2903036216 hasLocation W29030362162 @default.

• next