FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W4386869594> ?p ?o ?g. }

• W4386869594 endingPage "1" @default.
• W4386869594 startingPage "1" @default.
• W4386869594 abstract "There is a growing need to explore the potential of transformers in Unsupervised Domain Adaptation (UDA) due to their increasing success in various vision tasks. However, the application of transformers in UDA has yet to be thoroughly investigated and requires further research. In this study, our primary focus is to design a novel pipeline specifically tailored for transformer-based UDA, to address a crucial challenge: the overemphasis on the transfer of <italic xmlns:mml=http://www.w3.org/1998/Math/MathML xmlns:xlink=http://www.w3.org/1999/xlink>target</i> -oriented information, mainly caused by the self-attention blocks in transformers and the cross-domain adversarial learning scheme. First, we show that <italic xmlns:mml=http://www.w3.org/1998/Math/MathML xmlns:xlink=http://www.w3.org/1999/xlink>non-target</i> information, including semantic contextual information such as background features and <italic xmlns:mml=http://www.w3.org/1998/Math/MathML xmlns:xlink=http://www.w3.org/1999/xlink>non-target</i> classes, must be addressed in the domain adaptation process. Recognizing the importance of incorporating <italic xmlns:mml=http://www.w3.org/1998/Math/MathML xmlns:xlink=http://www.w3.org/1999/xlink>non-target</i> knowledge, we propose a decoupled <italic xmlns:mml=http://www.w3.org/1998/Math/MathML xmlns:xlink=http://www.w3.org/1999/xlink>non-target</i> knowledge distillation method called DeNKD. DeNKD decouples <italic xmlns:mml=http://www.w3.org/1998/Math/MathML xmlns:xlink=http://www.w3.org/1999/xlink>non-target</i> information across domains at both feature and logit levels. This decoupling is achieved through a bi-directional knowledge distillation approach that facilitates the interaction and exchange of <italic xmlns:mml=http://www.w3.org/1998/Math/MathML xmlns:xlink=http://www.w3.org/1999/xlink>non-target</i> knowledge to facilitate an effective transformer-based cross-domain knowledge transfer. We perform extensive evaluations on several well-established UDA benchmark datasets. The results consistently show that DeNKD outperforms other methods, achieving the best performance across the board. For example, on the Office-Home dataset, DeNKD achieves an accuracy of 85.54%, while on the VisDA-2017 dataset, it achieves an accuracy of 89.95%. These results highlight the effectiveness of DeNKD in transformer-based UDA and its potential for improving cross-domain adaptation performance." @default.
• W4386869594 created "2023-09-20" @default.
• W4386869594 creator A5021220108 @default.
• W4386869594 creator A5031432884 @default.
• W4386869594 creator A5049471074 @default.
• W4386869594 creator A5063499064 @default.
• W4386869594 creator A5081737485 @default.
• W4386869594 creator A5087818121 @default.
• W4386869594 date "2023-01-01" @default.
• W4386869594 modified "2023-09-27" @default.
• W4386869594 title "DeNKD: Decoupled <i>non-target</i> Knowledge Distillation for Complementing Transformer-based Unsupervised Domain Adaptation" @default.
• W4386869594 doi "https://doi.org/10.1109/tcsvt.2023.3315872" @default.
• W4386869594 hasPublicationYear "2023" @default.
• W4386869594 type Work @default.
• W4386869594 citedByCount "0" @default.
• W4386869594 crossrefType "journal-article" @default.
• W4386869594 hasAuthorship W4386869594A5021220108 @default.
• W4386869594 hasAuthorship W4386869594A5031432884 @default.
• W4386869594 hasAuthorship W4386869594A5049471074 @default.
• W4386869594 hasAuthorship W4386869594A5063499064 @default.
• W4386869594 hasAuthorship W4386869594A5081737485 @default.
• W4386869594 hasAuthorship W4386869594A5087818121 @default.
• W4386869594 hasConcept C119599485 @default.
• W4386869594 hasConcept C127413603 @default.
• W4386869594 hasConcept C134306372 @default.
• W4386869594 hasConcept C154945302 @default.
• W4386869594 hasConcept C165801399 @default.
• W4386869594 hasConcept C33923547 @default.
• W4386869594 hasConcept C36503486 @default.
• W4386869594 hasConcept C41008148 @default.
• W4386869594 hasConcept C66322947 @default.
• W4386869594 hasConceptScore W4386869594C119599485 @default.
• W4386869594 hasConceptScore W4386869594C127413603 @default.
• W4386869594 hasConceptScore W4386869594C134306372 @default.
• W4386869594 hasConceptScore W4386869594C154945302 @default.
• W4386869594 hasConceptScore W4386869594C165801399 @default.
• W4386869594 hasConceptScore W4386869594C33923547 @default.
• W4386869594 hasConceptScore W4386869594C36503486 @default.
• W4386869594 hasConceptScore W4386869594C41008148 @default.
• W4386869594 hasConceptScore W4386869594C66322947 @default.
• W4386869594 hasFunder F4320321001 @default.
• W4386869594 hasLocation W43868695941 @default.
• W4386869594 hasOpenAccess W4386869594 @default.
• W4386869594 hasPrimaryLocation W43868695941 @default.
• W4386869594 hasRelatedWork W1991466308 @default.
• W4386869594 hasRelatedWork W2358668433 @default.
• W4386869594 hasRelatedWork W2362192218 @default.
• W4386869594 hasRelatedWork W2390279801 @default.
• W4386869594 hasRelatedWork W2623347760 @default.
• W4386869594 hasRelatedWork W2743342830 @default.
• W4386869594 hasRelatedWork W2748952813 @default.
• W4386869594 hasRelatedWork W2899084033 @default.
• W4386869594 hasRelatedWork W3034138874 @default.
• W4386869594 hasRelatedWork W3157910026 @default.
• W4386869594 isParatext "false" @default.
• W4386869594 isRetracted "false" @default.
• W4386869594 workType "article" @default.