Matches in SemOpenAlex for { <https://semopenalex.org/work/W4386075596> ?p ?o ?g. }
Showing items 1 to 96 of
96
with 100 items per page.
- W4386075596 abstract "Visual information extraction (VIE) plays an important role in Document Intelligence. Generally, it is divided into two tasks: semantic entity recognition (SER) and relation extraction (RE). Recently, pre-trained models for documents have achieved substantial progress in VIE, particularly in SER. However, most of the existing models learn the geometric representation in an implicit way, which has been found insufficient for the RE task since geometric information is especially crucial for RE. Moreover, we reveal another factor that limits the performance of RE lies in the objective gap between the pre-training phase and the finetuning phase for RE. To tackle these issues, we propose in this paper a multi-modal framework, named GeoLayoutLM, for VIE. GeoLayoutLM explicitly models the geometric relations in pre-training, which we call geometric pre-training. Geometric pre-training is achieved by three specially designed geometry-related pre-training tasks. Additionally, novel relation heads, which are pre-trained by the geometric pre-training tasks and fine-tuned for RE, are elaborately designed to enrich and enhance the feature representation. According to extensive experiments on standard VIE benchmarks, GeoLayoutLM achieves highly competitive scores in the SER task and significantly outperforms the previous state-of-the-arts for RE (e.g., the F1 score of RE on FUNSD is boosted from 80.35% to 89.45%) <sup xmlns:mml=http://www.w3.org/1998/Math/MathML xmlns:xlink=http://www.w3.org/1999/xlink>1</sup> <sup xmlns:mml=http://www.w3.org/1998/Math/MathML xmlns:xlink=http://www.w3.org/1999/xlink>1</sup> https://github.com/AlibabaResearch/AdvancedLiterateMachinery." @default.
- W4386075596 created "2023-08-23" @default.
- W4386075596 creator A5057730293 @default.
- W4386075596 creator A5062254598 @default.
- W4386075596 creator A5069902262 @default.
- W4386075596 creator A5074055611 @default.
- W4386075596 date "2023-06-01" @default.
- W4386075596 modified "2023-09-27" @default.
- W4386075596 title "GeoLayoutLM: Geometric Pre-training for Visual Information Extraction" @default.
- W4386075596 cites W1966382373 @default.
- W4386075596 cites W2194187530 @default.
- W4386075596 cites W2605982830 @default.
- W4386075596 cites W2922714365 @default.
- W4386075596 cites W2927746189 @default.
- W4386075596 cites W2963026768 @default.
- W4386075596 cites W2963150697 @default.
- W4386075596 cites W2986619406 @default.
- W4386075596 cites W2998621280 @default.
- W4386075596 cites W3034238904 @default.
- W4386075596 cites W3082397598 @default.
- W4386075596 cites W3119740550 @default.
- W4386075596 cites W3170268576 @default.
- W4386075596 cites W3173306993 @default.
- W4386075596 cites W3173325518 @default.
- W4386075596 cites W3190448953 @default.
- W4386075596 cites W3200439183 @default.
- W4386075596 cites W3203055579 @default.
- W4386075596 cites W3205981739 @default.
- W4386075596 cites W3207806388 @default.
- W4386075596 cites W4221167941 @default.
- W4386075596 cites W4226020328 @default.
- W4386075596 cites W4229032688 @default.
- W4386075596 cites W4304013646 @default.
- W4386075596 cites W4312443924 @default.
- W4386075596 cites W4312784228 @default.
- W4386075596 cites W4312843595 @default.
- W4386075596 cites W654550266 @default.
- W4386075596 doi "https://doi.org/10.1109/cvpr52729.2023.00685" @default.
- W4386075596 hasPublicationYear "2023" @default.
- W4386075596 type Work @default.
- W4386075596 citedByCount "0" @default.
- W4386075596 crossrefType "proceedings-article" @default.
- W4386075596 hasAuthorship W4386075596A5057730293 @default.
- W4386075596 hasAuthorship W4386075596A5062254598 @default.
- W4386075596 hasAuthorship W4386075596A5069902262 @default.
- W4386075596 hasAuthorship W4386075596A5074055611 @default.
- W4386075596 hasConcept C119857082 @default.
- W4386075596 hasConcept C124101348 @default.
- W4386075596 hasConcept C153180895 @default.
- W4386075596 hasConcept C153604712 @default.
- W4386075596 hasConcept C154945302 @default.
- W4386075596 hasConcept C162324750 @default.
- W4386075596 hasConcept C17744445 @default.
- W4386075596 hasConcept C187736073 @default.
- W4386075596 hasConcept C195807954 @default.
- W4386075596 hasConcept C199539241 @default.
- W4386075596 hasConcept C204321447 @default.
- W4386075596 hasConcept C25343380 @default.
- W4386075596 hasConcept C2776359362 @default.
- W4386075596 hasConcept C2780451532 @default.
- W4386075596 hasConcept C41008148 @default.
- W4386075596 hasConcept C52622490 @default.
- W4386075596 hasConcept C94625758 @default.
- W4386075596 hasConceptScore W4386075596C119857082 @default.
- W4386075596 hasConceptScore W4386075596C124101348 @default.
- W4386075596 hasConceptScore W4386075596C153180895 @default.
- W4386075596 hasConceptScore W4386075596C153604712 @default.
- W4386075596 hasConceptScore W4386075596C154945302 @default.
- W4386075596 hasConceptScore W4386075596C162324750 @default.
- W4386075596 hasConceptScore W4386075596C17744445 @default.
- W4386075596 hasConceptScore W4386075596C187736073 @default.
- W4386075596 hasConceptScore W4386075596C195807954 @default.
- W4386075596 hasConceptScore W4386075596C199539241 @default.
- W4386075596 hasConceptScore W4386075596C204321447 @default.
- W4386075596 hasConceptScore W4386075596C25343380 @default.
- W4386075596 hasConceptScore W4386075596C2776359362 @default.
- W4386075596 hasConceptScore W4386075596C2780451532 @default.
- W4386075596 hasConceptScore W4386075596C41008148 @default.
- W4386075596 hasConceptScore W4386075596C52622490 @default.
- W4386075596 hasConceptScore W4386075596C94625758 @default.
- W4386075596 hasLocation W43860755961 @default.
- W4386075596 hasOpenAccess W4386075596 @default.
- W4386075596 hasPrimaryLocation W43860755961 @default.
- W4386075596 hasRelatedWork W2146076056 @default.
- W4386075596 hasRelatedWork W2753023842 @default.
- W4386075596 hasRelatedWork W2807524541 @default.
- W4386075596 hasRelatedWork W2811390910 @default.
- W4386075596 hasRelatedWork W2888033806 @default.
- W4386075596 hasRelatedWork W3114114934 @default.
- W4386075596 hasRelatedWork W4224089748 @default.
- W4386075596 hasRelatedWork W4299912061 @default.
- W4386075596 hasRelatedWork W4319940250 @default.
- W4386075596 hasRelatedWork W4379744446 @default.
- W4386075596 isParatext "false" @default.
- W4386075596 isRetracted "false" @default.
- W4386075596 workType "article" @default.