Matches in SemOpenAlex for { <https://semopenalex.org/work/W3157643567> ?p ?o ?g. }
- W3157643567 abstract "Unsupervised image captioning is a challenging task that aims at generating captions without the supervision of image-sentence pairs, but only with images and sentences drawn from different sources and object labels detected from the images. In previous work, pseudo-captions, i.e., sentences that contain the detected object labels, were assigned to a given image. The focus of the previous work was on the alignment of input images and pseudo-captions at the sentence level. However, pseudo-captions contain many words that are irrelevant to a given image. In this work, we investigate the effect of removing mismatched words from image-sentence alignment to determine how they make this task difficult. We propose a simple gating mechanism that is trained to align image features with only the most reliable words in pseudo-captions: the detected object labels. The experimental results show that our proposed method outperforms the previous methods without introducing complex sentence-level learning objectives. Combined with the sentence-level alignment method of previous work, our method further improves its performance. These results confirm the importance of careful alignment in word-level details." @default.
- W3157643567 created "2021-05-10" @default.
- W3157643567 creator A5038408644 @default.
- W3157643567 creator A5071438625 @default.
- W3157643567 creator A5072032804 @default.
- W3157643567 creator A5077707500 @default.
- W3157643567 creator A5081626471 @default.
- W3157643567 date "2021-04-28" @default.
- W3157643567 modified "2023-10-12" @default.
- W3157643567 title "Removing Word-Level Spurious Alignment between Images and Pseudo-Captions in Unsupervised Image Captioning" @default.
- W3157643567 cites W1514535095 @default.
- W3157643567 cites W1861492603 @default.
- W3157643567 cites W1895577753 @default.
- W3157643567 cites W1956340063 @default.
- W3157643567 cites W2064675550 @default.
- W3157643567 cites W2099471712 @default.
- W3157643567 cites W2101105183 @default.
- W3157643567 cites W2117539524 @default.
- W3157643567 cites W2133459682 @default.
- W3157643567 cites W2154652894 @default.
- W3157643567 cites W2157331557 @default.
- W3157643567 cites W2173180041 @default.
- W3157643567 cites W2185175083 @default.
- W3157643567 cites W2194775991 @default.
- W3157643567 cites W2277195237 @default.
- W3157643567 cites W2302255633 @default.
- W3157643567 cites W2506483933 @default.
- W3157643567 cites W2557728737 @default.
- W3157643567 cites W2575842049 @default.
- W3157643567 cites W2613718673 @default.
- W3157643567 cites W2745461083 @default.
- W3157643567 cites W2795151422 @default.
- W3157643567 cites W2886641317 @default.
- W3157643567 cites W2950178297 @default.
- W3157643567 cites W2963048642 @default.
- W3157643567 cites W2963049308 @default.
- W3157643567 cites W2963088515 @default.
- W3157643567 cites W2963170456 @default.
- W3157643567 cites W2963743213 @default.
- W3157643567 cites W2964121744 @default.
- W3157643567 cites W2964350391 @default.
- W3157643567 cites W2981994675 @default.
- W3157643567 cites W2990307191 @default.
- W3157643567 cites W2992478697 @default.
- W3157643567 cites W3004328990 @default.
- W3157643567 cites W3034493371 @default.
- W3157643567 cites W3086385316 @default.
- W3157643567 cites W3102877762 @default.
- W3157643567 cites W3103211586 @default.
- W3157643567 cites W3104279398 @default.
- W3157643567 cites W3121480429 @default.
- W3157643567 doi "https://doi.org/10.48550/arxiv.2104.13872" @default.
- W3157643567 hasPublicationYear "2021" @default.
- W3157643567 type Work @default.
- W3157643567 sameAs 3157643567 @default.
- W3157643567 citedByCount "1" @default.
- W3157643567 countsByYear W31576435672021 @default.
- W3157643567 crossrefType "posted-content" @default.
- W3157643567 hasAuthorship W3157643567A5038408644 @default.
- W3157643567 hasAuthorship W3157643567A5071438625 @default.
- W3157643567 hasAuthorship W3157643567A5072032804 @default.
- W3157643567 hasAuthorship W3157643567A5077707500 @default.
- W3157643567 hasAuthorship W3157643567A5081626471 @default.
- W3157643567 hasBestOaLocation W31576435671 @default.
- W3157643567 hasConcept C115961682 @default.
- W3157643567 hasConcept C119857082 @default.
- W3157643567 hasConcept C120665830 @default.
- W3157643567 hasConcept C121332964 @default.
- W3157643567 hasConcept C138885662 @default.
- W3157643567 hasConcept C153180895 @default.
- W3157643567 hasConcept C154945302 @default.
- W3157643567 hasConcept C157657479 @default.
- W3157643567 hasConcept C162324750 @default.
- W3157643567 hasConcept C187736073 @default.
- W3157643567 hasConcept C192209626 @default.
- W3157643567 hasConcept C204321447 @default.
- W3157643567 hasConcept C2777530160 @default.
- W3157643567 hasConcept C2780451532 @default.
- W3157643567 hasConcept C2781238097 @default.
- W3157643567 hasConcept C28490314 @default.
- W3157643567 hasConcept C31972630 @default.
- W3157643567 hasConcept C41008148 @default.
- W3157643567 hasConcept C41895202 @default.
- W3157643567 hasConcept C90805587 @default.
- W3157643567 hasConcept C97256817 @default.
- W3157643567 hasConceptScore W3157643567C115961682 @default.
- W3157643567 hasConceptScore W3157643567C119857082 @default.
- W3157643567 hasConceptScore W3157643567C120665830 @default.
- W3157643567 hasConceptScore W3157643567C121332964 @default.
- W3157643567 hasConceptScore W3157643567C138885662 @default.
- W3157643567 hasConceptScore W3157643567C153180895 @default.
- W3157643567 hasConceptScore W3157643567C154945302 @default.
- W3157643567 hasConceptScore W3157643567C157657479 @default.
- W3157643567 hasConceptScore W3157643567C162324750 @default.
- W3157643567 hasConceptScore W3157643567C187736073 @default.
- W3157643567 hasConceptScore W3157643567C192209626 @default.
- W3157643567 hasConceptScore W3157643567C204321447 @default.
- W3157643567 hasConceptScore W3157643567C2777530160 @default.
- W3157643567 hasConceptScore W3157643567C2780451532 @default.
- W3157643567 hasConceptScore W3157643567C2781238097 @default.