Matches in SemOpenAlex for { <https://semopenalex.org/work/W4301186750> ?p ?o ?g. }
Showing items 1 to 91 of
91
with 100 items per page.
- W4301186750 abstract "The adoption of deep learning models has brought significant performance improvements across several research fields, such as computer vision and natural language processing. However, their “black-box” nature yields the downside of poor explainability: in particular, several real-world applications require - to varying extents - reliable confidence scores associated to a model's prediction. The relation between a model's accuracy and confidence is typically referred to as calibration. In this work, we propose a novel calibration method based on gradient accumulation in conjunction with existing loss regularization techniques. Our experiments on the Named Entity Recognition task show an improvement of the performance/calibration ratio compared to the current methods." @default.
- W4301186750 created "2022-10-04" @default.
- W4301186750 creator A5032407315 @default.
- W4301186750 creator A5038438830 @default.
- W4301186750 creator A5045096979 @default.
- W4301186750 creator A5082391569 @default.
- W4301186750 creator A5085150504 @default.
- W4301186750 date "2022-07-18" @default.
- W4301186750 modified "2023-09-27" @default.
- W4301186750 title "A Novel Gradient Accumulation Method for Calibration of Named Entity Recognition Models" @default.
- W4301186750 cites W1540419678 @default.
- W4301186750 cites W2030214786 @default.
- W4301186750 cites W2047634553 @default.
- W4301186750 cites W2098057544 @default.
- W4301186750 cites W2098700435 @default.
- W4301186750 cites W2098824882 @default.
- W4301186750 cites W2169099542 @default.
- W4301186750 cites W2299976354 @default.
- W4301186750 cites W2742113707 @default.
- W4301186750 cites W2759245808 @default.
- W4301186750 cites W2954996726 @default.
- W4301186750 cites W2963266575 @default.
- W4301186750 cites W2963485691 @default.
- W4301186750 cites W2979826702 @default.
- W4301186750 cites W2982509564 @default.
- W4301186750 cites W3021332602 @default.
- W4301186750 cites W3034704352 @default.
- W4301186750 cites W3104939451 @default.
- W4301186750 cites W3106938322 @default.
- W4301186750 cites W3173618889 @default.
- W4301186750 cites W3174828871 @default.
- W4301186750 cites W3199958362 @default.
- W4301186750 cites W3205854703 @default.
- W4301186750 cites W4205373355 @default.
- W4301186750 doi "https://doi.org/10.1109/ijcnn55064.2022.9892324" @default.
- W4301186750 hasPublicationYear "2022" @default.
- W4301186750 type Work @default.
- W4301186750 citedByCount "0" @default.
- W4301186750 crossrefType "proceedings-article" @default.
- W4301186750 hasAuthorship W4301186750A5032407315 @default.
- W4301186750 hasAuthorship W4301186750A5038438830 @default.
- W4301186750 hasAuthorship W4301186750A5045096979 @default.
- W4301186750 hasAuthorship W4301186750A5082391569 @default.
- W4301186750 hasAuthorship W4301186750A5085150504 @default.
- W4301186750 hasBestOaLocation W43011867502 @default.
- W4301186750 hasConcept C105795698 @default.
- W4301186750 hasConcept C119857082 @default.
- W4301186750 hasConcept C124101348 @default.
- W4301186750 hasConcept C127413603 @default.
- W4301186750 hasConcept C153180895 @default.
- W4301186750 hasConcept C154945302 @default.
- W4301186750 hasConcept C165838908 @default.
- W4301186750 hasConcept C201995342 @default.
- W4301186750 hasConcept C204321447 @default.
- W4301186750 hasConcept C25343380 @default.
- W4301186750 hasConcept C2776135515 @default.
- W4301186750 hasConcept C2780451532 @default.
- W4301186750 hasConcept C33923547 @default.
- W4301186750 hasConcept C41008148 @default.
- W4301186750 hasConceptScore W4301186750C105795698 @default.
- W4301186750 hasConceptScore W4301186750C119857082 @default.
- W4301186750 hasConceptScore W4301186750C124101348 @default.
- W4301186750 hasConceptScore W4301186750C127413603 @default.
- W4301186750 hasConceptScore W4301186750C153180895 @default.
- W4301186750 hasConceptScore W4301186750C154945302 @default.
- W4301186750 hasConceptScore W4301186750C165838908 @default.
- W4301186750 hasConceptScore W4301186750C201995342 @default.
- W4301186750 hasConceptScore W4301186750C204321447 @default.
- W4301186750 hasConceptScore W4301186750C25343380 @default.
- W4301186750 hasConceptScore W4301186750C2776135515 @default.
- W4301186750 hasConceptScore W4301186750C2780451532 @default.
- W4301186750 hasConceptScore W4301186750C33923547 @default.
- W4301186750 hasConceptScore W4301186750C41008148 @default.
- W4301186750 hasLocation W43011867501 @default.
- W4301186750 hasLocation W43011867502 @default.
- W4301186750 hasLocation W43011867503 @default.
- W4301186750 hasOpenAccess W4301186750 @default.
- W4301186750 hasPrimaryLocation W43011867501 @default.
- W4301186750 hasRelatedWork W2045408812 @default.
- W4301186750 hasRelatedWork W2081647779 @default.
- W4301186750 hasRelatedWork W2961085424 @default.
- W4301186750 hasRelatedWork W3046775127 @default.
- W4301186750 hasRelatedWork W3170094116 @default.
- W4301186750 hasRelatedWork W4285260836 @default.
- W4301186750 hasRelatedWork W4286629047 @default.
- W4301186750 hasRelatedWork W4306321456 @default.
- W4301186750 hasRelatedWork W4306674287 @default.
- W4301186750 hasRelatedWork W4224009465 @default.
- W4301186750 isParatext "false" @default.
- W4301186750 isRetracted "false" @default.
- W4301186750 workType "article" @default.