Matches in SemOpenAlex for { <https://semopenalex.org/work/W3204254991> ?p ?o ?g. }
Showing items 1 to 97 of
97
with 100 items per page.
- W3204254991 abstract "Documents are central to many business systems, and include forms, reports, contracts, invoices or purchase orders. The information in documents is typically in natural language, but can be organized in various layouts and formats. There have been recent spurt of interest in understanding document content with novel deep learning architectures. However, document understanding tasks need dense information annotations, which are costly to scale and generalize. Several active learning techniques have been proposed to reduce the overall budget of annotation while maintaining the performance of the underlying deep learning model. In this paper, we propose OPAD, a novel framework using reinforcement policy for active learning in content detection tasks for documents. The proposed framework learns the acquisition function to decide the samples to be selected while optimizing performance metrics that the tasks typically have. Furthermore, we extend to weak labelling scenarios to further reduce the cost of annotation significantly. We propose novel rewards to account for class imbalance and user feedback in the annotation interface, to improve the active learning method. We show superior performance of the proposed OPAD framework for active learning for various tasks related to document understanding like layout parsing, object detection and named entity recognition. Ablation studies for human feedback and class imbalance rewards are presented, along with a comparison of annotation times for different approaches." @default.
- W3204254991 created "2021-10-11" @default.
- W3204254991 creator A5006519552 @default.
- W3204254991 creator A5010465342 @default.
- W3204254991 creator A5045320432 @default.
- W3204254991 creator A5077046363 @default.
- W3204254991 creator A5085015139 @default.
- W3204254991 date "2022-06-01" @default.
- W3204254991 modified "2023-10-16" @default.
- W3204254991 title "OPAD: An Optimized Policy-based Active Learning Framework for Document Content Analysis" @default.
- W3204254991 cites W1777939696 @default.
- W3204254991 cites W1966382373 @default.
- W3204254991 cites W1997754546 @default.
- W3204254991 cites W2011674654 @default.
- W3204254991 cites W2061688106 @default.
- W3204254991 cites W2262342046 @default.
- W3204254991 cites W2471138382 @default.
- W3204254991 cites W2608567823 @default.
- W3204254991 cites W2769387903 @default.
- W3204254991 cites W2785787385 @default.
- W3204254991 cites W2798587560 @default.
- W3204254991 cites W2899181341 @default.
- W3204254991 cites W2956371155 @default.
- W3204254991 cites W2962772269 @default.
- W3204254991 cites W2963696295 @default.
- W3204254991 cites W2964073328 @default.
- W3204254991 cites W2964206023 @default.
- W3204254991 cites W2964340375 @default.
- W3204254991 cites W2965018299 @default.
- W3204254991 cites W2978725006 @default.
- W3204254991 cites W2980913015 @default.
- W3204254991 cites W2983697263 @default.
- W3204254991 cites W2984303785 @default.
- W3204254991 cites W2998936706 @default.
- W3204254991 cites W3003273206 @default.
- W3204254991 cites W3003711898 @default.
- W3204254991 cites W3008736151 @default.
- W3204254991 cites W3025091470 @default.
- W3204254991 cites W3034942609 @default.
- W3204254991 cites W3039883906 @default.
- W3204254991 cites W3041202696 @default.
- W3204254991 cites W3098052913 @default.
- W3204254991 cites W3104953317 @default.
- W3204254991 cites W3105988348 @default.
- W3204254991 cites W3113753692 @default.
- W3204254991 cites W3204999291 @default.
- W3204254991 doi "https://doi.org/10.1109/cvprw56347.2022.00320" @default.
- W3204254991 hasPublicationYear "2022" @default.
- W3204254991 type Work @default.
- W3204254991 sameAs 3204254991 @default.
- W3204254991 citedByCount "1" @default.
- W3204254991 countsByYear W32042549912023 @default.
- W3204254991 crossrefType "proceedings-article" @default.
- W3204254991 hasAuthorship W3204254991A5006519552 @default.
- W3204254991 hasAuthorship W3204254991A5010465342 @default.
- W3204254991 hasAuthorship W3204254991A5045320432 @default.
- W3204254991 hasAuthorship W3204254991A5077046363 @default.
- W3204254991 hasAuthorship W3204254991A5085015139 @default.
- W3204254991 hasBestOaLocation W32042549912 @default.
- W3204254991 hasConcept C119857082 @default.
- W3204254991 hasConcept C154945302 @default.
- W3204254991 hasConcept C186644900 @default.
- W3204254991 hasConcept C204321447 @default.
- W3204254991 hasConcept C23123220 @default.
- W3204254991 hasConcept C2776321320 @default.
- W3204254991 hasConcept C2777212361 @default.
- W3204254991 hasConcept C41008148 @default.
- W3204254991 hasConcept C77967617 @default.
- W3204254991 hasConcept C97541855 @default.
- W3204254991 hasConceptScore W3204254991C119857082 @default.
- W3204254991 hasConceptScore W3204254991C154945302 @default.
- W3204254991 hasConceptScore W3204254991C186644900 @default.
- W3204254991 hasConceptScore W3204254991C204321447 @default.
- W3204254991 hasConceptScore W3204254991C23123220 @default.
- W3204254991 hasConceptScore W3204254991C2776321320 @default.
- W3204254991 hasConceptScore W3204254991C2777212361 @default.
- W3204254991 hasConceptScore W3204254991C41008148 @default.
- W3204254991 hasConceptScore W3204254991C77967617 @default.
- W3204254991 hasConceptScore W3204254991C97541855 @default.
- W3204254991 hasLocation W32042549911 @default.
- W3204254991 hasLocation W32042549912 @default.
- W3204254991 hasOpenAccess W3204254991 @default.
- W3204254991 hasPrimaryLocation W32042549911 @default.
- W3204254991 hasRelatedWork W1533177136 @default.
- W3204254991 hasRelatedWork W2053943328 @default.
- W3204254991 hasRelatedWork W2218034408 @default.
- W3204254991 hasRelatedWork W2263699433 @default.
- W3204254991 hasRelatedWork W2358755282 @default.
- W3204254991 hasRelatedWork W2361861616 @default.
- W3204254991 hasRelatedWork W2377979023 @default.
- W3204254991 hasRelatedWork W2392921965 @default.
- W3204254991 hasRelatedWork W2625833328 @default.
- W3204254991 hasRelatedWork W3046984657 @default.
- W3204254991 isParatext "false" @default.
- W3204254991 isRetracted "false" @default.
- W3204254991 magId "3204254991" @default.
- W3204254991 workType "article" @default.