Matches in SemOpenAlex for { <https://semopenalex.org/work/W2972025845> ?p ?o ?g. }
- W2972025845 abstract "Task-specific scores are often used to optimize for and evaluate the performance of conditional text generation systems. However, such scores are non-differentiable and cannot be used in the standard supervised learning paradigm. Hence, policy gradient methods are used since the gradient can be computed without requiring a differentiable objective. However, we argue that current n-gram overlap based measures that are used as rewards can be improved by using model-based rewards transferred from tasks that directly compare the similarity of sentence pairs. These reward models either output a score of sentence-level syntactic and semantic similarity between entire predicted and target sentences as the expected return, or for intermediate phrases as segmented accumulative rewards. We demonstrate that using a textit{Transferable Reward Learner} leads to improved results on semantical evaluation measures in policy-gradient models for image captioning tasks. Our InferSent actor-critic model improves over a BLEU trained actor-critic model on MSCOCO when evaluated on a Word Mover's Distance similarity measure by 6.97 points, also improving on a Sliding Window Cosine Similarity measure by 10.48 points. Similar performance improvements are also obtained on the smaller Flickr-30k dataset, demonstrating the general applicability of the proposed transfer learning method." @default.
- W2972025845 created "2019-09-12" @default.
- W2972025845 creator A5073503574 @default.
- W2972025845 creator A5083742715 @default.
- W2972025845 date "2019-09-09" @default.
- W2972025845 modified "2023-10-17" @default.
- W2972025845 title "Transfer Reward Learning for Policy Gradient-Based Text Generation." @default.
- W2972025845 cites W112666333 @default.
- W2972025845 cites W1514535095 @default.
- W2972025845 cites W1522301498 @default.
- W2972025845 cites W1895577753 @default.
- W2972025845 cites W1895989618 @default.
- W2972025845 cites W1905882502 @default.
- W2972025845 cites W1947481528 @default.
- W2972025845 cites W1956340063 @default.
- W2972025845 cites W1999874108 @default.
- W2972025845 cites W2101105183 @default.
- W2972025845 cites W2119717200 @default.
- W2972025845 cites W2133459682 @default.
- W2972025845 cites W2155027007 @default.
- W2972025845 cites W2176263492 @default.
- W2972025845 cites W2194775991 @default.
- W2972025845 cites W2274287116 @default.
- W2972025845 cites W2410323755 @default.
- W2972025845 cites W2487501366 @default.
- W2972025845 cites W2558533273 @default.
- W2972025845 cites W2561386235 @default.
- W2972025845 cites W2618606525 @default.
- W2972025845 cites W2626804490 @default.
- W2972025845 cites W2729842244 @default.
- W2972025845 cites W2950178297 @default.
- W2972025845 cites W2950726992 @default.
- W2972025845 cites W2952591111 @default.
- W2972025845 cites W2953084091 @default.
- W2972025845 cites W2962706528 @default.
- W2972025845 cites W2963084599 @default.
- W2972025845 cites W2963341956 @default.
- W2972025845 cites W2963630207 @default.
- W2972025845 cites W2963962369 @default.
- W2972025845 cites W2964042428 @default.
- W2972025845 cites W658020064 @default.
- W2972025845 hasPublicationYear "2019" @default.
- W2972025845 type Work @default.
- W2972025845 sameAs 2972025845 @default.
- W2972025845 citedByCount "1" @default.
- W2972025845 countsByYear W29720258452022 @default.
- W2972025845 crossrefType "posted-content" @default.
- W2972025845 hasAuthorship W2972025845A5073503574 @default.
- W2972025845 hasAuthorship W2972025845A5083742715 @default.
- W2972025845 hasConcept C103278499 @default.
- W2972025845 hasConcept C115961682 @default.
- W2972025845 hasConcept C119857082 @default.
- W2972025845 hasConcept C124101348 @default.
- W2972025845 hasConcept C134306372 @default.
- W2972025845 hasConcept C150899416 @default.
- W2972025845 hasConcept C153180895 @default.
- W2972025845 hasConcept C154945302 @default.
- W2972025845 hasConcept C157657479 @default.
- W2972025845 hasConcept C162324750 @default.
- W2972025845 hasConcept C187736073 @default.
- W2972025845 hasConcept C202615002 @default.
- W2972025845 hasConcept C204321447 @default.
- W2972025845 hasConcept C2524010 @default.
- W2972025845 hasConcept C2777530160 @default.
- W2972025845 hasConcept C2780009758 @default.
- W2972025845 hasConcept C2780451532 @default.
- W2972025845 hasConcept C2780762811 @default.
- W2972025845 hasConcept C33923547 @default.
- W2972025845 hasConcept C41008148 @default.
- W2972025845 hasConcept C90805587 @default.
- W2972025845 hasConceptScore W2972025845C103278499 @default.
- W2972025845 hasConceptScore W2972025845C115961682 @default.
- W2972025845 hasConceptScore W2972025845C119857082 @default.
- W2972025845 hasConceptScore W2972025845C124101348 @default.
- W2972025845 hasConceptScore W2972025845C134306372 @default.
- W2972025845 hasConceptScore W2972025845C150899416 @default.
- W2972025845 hasConceptScore W2972025845C153180895 @default.
- W2972025845 hasConceptScore W2972025845C154945302 @default.
- W2972025845 hasConceptScore W2972025845C157657479 @default.
- W2972025845 hasConceptScore W2972025845C162324750 @default.
- W2972025845 hasConceptScore W2972025845C187736073 @default.
- W2972025845 hasConceptScore W2972025845C202615002 @default.
- W2972025845 hasConceptScore W2972025845C204321447 @default.
- W2972025845 hasConceptScore W2972025845C2524010 @default.
- W2972025845 hasConceptScore W2972025845C2777530160 @default.
- W2972025845 hasConceptScore W2972025845C2780009758 @default.
- W2972025845 hasConceptScore W2972025845C2780451532 @default.
- W2972025845 hasConceptScore W2972025845C2780762811 @default.
- W2972025845 hasConceptScore W2972025845C33923547 @default.
- W2972025845 hasConceptScore W2972025845C41008148 @default.
- W2972025845 hasConceptScore W2972025845C90805587 @default.
- W2972025845 hasLocation W29720258451 @default.
- W2972025845 hasOpenAccess W2972025845 @default.
- W2972025845 hasPrimaryLocation W29720258451 @default.
- W2972025845 hasRelatedWork W1950554513 @default.
- W2972025845 hasRelatedWork W2139725124 @default.
- W2972025845 hasRelatedWork W2252025371 @default.
- W2972025845 hasRelatedWork W2541478194 @default.
- W2972025845 hasRelatedWork W2764124976 @default.
- W2972025845 hasRelatedWork W2768029001 @default.