Matches in SemOpenAlex for { <https://semopenalex.org/work/W4312852287> ?p ?o ?g. }
Showing items 1 to 88 of
88
with 100 items per page.
- W4312852287 abstract "The Transformer is one of the mainstream methods in computer vision. Most Transformer based architectures focus on the design of spatial attention and optimizing the computational complexity of high resolution of pixels in image but pay little attention to modeling channel dependencies and optimizing the computational complexity associated with a large number of channels. In this paper, we propose a new channel window based self-attention mechanism and apply two consecutive transformer layers to capture global channel dependencies through permuting channel layer, which can greatly reduce the computational complexity caused by a large number of channels. Meanwhile, a new linear layer for channel attention is proposed, which eliminates the need for position bias in Transformer. The proposed method can be conveniently appended to the existing image classification architectures in parallel with minimal modification. We demonstrate the feasibility of the proposed method on the state-of-the-art transformer-based architecture for image classification and improve the results on ImageNet-1K. The code will be publicly available at GitHub." @default.
- W4312852287 created "2023-01-05" @default.
- W4312852287 creator A5013579299 @default.
- W4312852287 creator A5029595402 @default.
- W4312852287 creator A5036622666 @default.
- W4312852287 creator A5038500322 @default.
- W4312852287 creator A5043869758 @default.
- W4312852287 creator A5060886777 @default.
- W4312852287 creator A5071740301 @default.
- W4312852287 date "2022-07-18" @default.
- W4312852287 modified "2023-09-23" @default.
- W4312852287 title "Pcwin Transformer: Permuted Channel Window based Attention for Image Classification" @default.
- W4312852287 cites W1861492603 @default.
- W4312852287 cites W2097117768 @default.
- W4312852287 cites W2108598243 @default.
- W4312852287 cites W2194775991 @default.
- W4312852287 cites W2549139847 @default.
- W4312852287 cites W2601564443 @default.
- W4312852287 cites W2963091558 @default.
- W4312852287 cites W2963420686 @default.
- W4312852287 cites W2963446712 @default.
- W4312852287 cites W2964231884 @default.
- W4312852287 cites W2966926453 @default.
- W4312852287 cites W2982220924 @default.
- W4312852287 cites W2983446232 @default.
- W4312852287 cites W3014641072 @default.
- W4312852287 cites W3034429256 @default.
- W4312852287 cites W3034552520 @default.
- W4312852287 cites W3131500599 @default.
- W4312852287 cites W3138516171 @default.
- W4312852287 cites W3172509117 @default.
- W4312852287 cites W3177052299 @default.
- W4312852287 doi "https://doi.org/10.1109/ijcnn55064.2022.9892630" @default.
- W4312852287 hasPublicationYear "2022" @default.
- W4312852287 type Work @default.
- W4312852287 citedByCount "0" @default.
- W4312852287 crossrefType "proceedings-article" @default.
- W4312852287 hasAuthorship W4312852287A5013579299 @default.
- W4312852287 hasAuthorship W4312852287A5029595402 @default.
- W4312852287 hasAuthorship W4312852287A5036622666 @default.
- W4312852287 hasAuthorship W4312852287A5038500322 @default.
- W4312852287 hasAuthorship W4312852287A5043869758 @default.
- W4312852287 hasAuthorship W4312852287A5060886777 @default.
- W4312852287 hasAuthorship W4312852287A5071740301 @default.
- W4312852287 hasConcept C113775141 @default.
- W4312852287 hasConcept C11413529 @default.
- W4312852287 hasConcept C119599485 @default.
- W4312852287 hasConcept C123657996 @default.
- W4312852287 hasConcept C127413603 @default.
- W4312852287 hasConcept C142362112 @default.
- W4312852287 hasConcept C153349607 @default.
- W4312852287 hasConcept C154945302 @default.
- W4312852287 hasConcept C160633673 @default.
- W4312852287 hasConcept C165801399 @default.
- W4312852287 hasConcept C179799912 @default.
- W4312852287 hasConcept C41008148 @default.
- W4312852287 hasConcept C66322947 @default.
- W4312852287 hasConceptScore W4312852287C113775141 @default.
- W4312852287 hasConceptScore W4312852287C11413529 @default.
- W4312852287 hasConceptScore W4312852287C119599485 @default.
- W4312852287 hasConceptScore W4312852287C123657996 @default.
- W4312852287 hasConceptScore W4312852287C127413603 @default.
- W4312852287 hasConceptScore W4312852287C142362112 @default.
- W4312852287 hasConceptScore W4312852287C153349607 @default.
- W4312852287 hasConceptScore W4312852287C154945302 @default.
- W4312852287 hasConceptScore W4312852287C160633673 @default.
- W4312852287 hasConceptScore W4312852287C165801399 @default.
- W4312852287 hasConceptScore W4312852287C179799912 @default.
- W4312852287 hasConceptScore W4312852287C41008148 @default.
- W4312852287 hasConceptScore W4312852287C66322947 @default.
- W4312852287 hasFunder F4320321001 @default.
- W4312852287 hasFunder F4320324174 @default.
- W4312852287 hasLocation W43128522871 @default.
- W4312852287 hasOpenAccess W4312852287 @default.
- W4312852287 hasPrimaryLocation W43128522871 @default.
- W4312852287 hasRelatedWork W2090093270 @default.
- W4312852287 hasRelatedWork W2114889067 @default.
- W4312852287 hasRelatedWork W2135989223 @default.
- W4312852287 hasRelatedWork W2364531466 @default.
- W4312852287 hasRelatedWork W2383138225 @default.
- W4312852287 hasRelatedWork W2383828164 @default.
- W4312852287 hasRelatedWork W2719428933 @default.
- W4312852287 hasRelatedWork W3107474891 @default.
- W4312852287 hasRelatedWork W3115043162 @default.
- W4312852287 hasRelatedWork W4312852287 @default.
- W4312852287 isParatext "false" @default.
- W4312852287 isRetracted "false" @default.
- W4312852287 workType "article" @default.