Matches in SemOpenAlex for { <https://semopenalex.org/work/W4386072014> ?p ?o ?g. }
Showing items 1 to 67 of
67
with 100 items per page.
- W4386072014 abstract "Recently vision transformer models have become prominent models for a range of tasks. These models, however, usually suffer from intensive computational costs and heavy memory requirements, making them impractical for deployment on edge platforms. Recent studies have proposed to prune transformers in an unexplainable manner, which overlook the relationship between internal units of the model and the target class, thereby leading to inferior performance. To alleviate this problem, we propose a novel explainable pruning framework dubbed X-Pruner, which is designed by considering the explainability of the pruning criterion. Specifically, to measure each prunable unit's contribution to predicting each target class, a novel explainability-aware mask is proposed and learned in an end-to-end manner. Then, to preserve the most informative units and learn the layer-wise pruning rate, we adaptively search the layer-wise threshold that differentiates between unpruned and pruned units based on their explainability-aware mask values. To verify and evaluate our method, we apply the X-Pruner on representative transformer models including the DeiT and Swin Transformer. Comprehensive simulation results demonstrate that the proposed X-Pruner outperforms the state-of-the-art black-box methods with significantly reduced computational costs and slight performance degradation. Code is available at https://github.com/vickyyu90/XPruner." @default.
- W4386072014 created "2023-08-23" @default.
- W4386072014 creator A5035064408 @default.
- W4386072014 creator A5044936528 @default.
- W4386072014 date "2023-06-01" @default.
- W4386072014 modified "2023-10-16" @default.
- W4386072014 title "X-Pruner: eXplainable Pruning for Vision Transformers" @default.
- W4386072014 cites W2108598243 @default.
- W4386072014 cites W2948947170 @default.
- W4386072014 cites W2981731882 @default.
- W4386072014 cites W2986670728 @default.
- W4386072014 cites W3035422918 @default.
- W4386072014 cites W3104216863 @default.
- W4386072014 cites W3138516171 @default.
- W4386072014 cites W3174173633 @default.
- W4386072014 cites W3176196997 @default.
- W4386072014 cites W3192814580 @default.
- W4386072014 cites W4206281850 @default.
- W4386072014 cites W4214633470 @default.
- W4386072014 cites W4230405732 @default.
- W4386072014 cites W4283811336 @default.
- W4386072014 cites W4312772337 @default.
- W4386072014 doi "https://doi.org/10.1109/cvpr52729.2023.02333" @default.
- W4386072014 hasPublicationYear "2023" @default.
- W4386072014 type Work @default.
- W4386072014 citedByCount "0" @default.
- W4386072014 crossrefType "proceedings-article" @default.
- W4386072014 hasAuthorship W4386072014A5035064408 @default.
- W4386072014 hasAuthorship W4386072014A5044936528 @default.
- W4386072014 hasConcept C108010975 @default.
- W4386072014 hasConcept C113775141 @default.
- W4386072014 hasConcept C119599485 @default.
- W4386072014 hasConcept C119857082 @default.
- W4386072014 hasConcept C127413603 @default.
- W4386072014 hasConcept C154945302 @default.
- W4386072014 hasConcept C165801399 @default.
- W4386072014 hasConcept C41008148 @default.
- W4386072014 hasConcept C6557445 @default.
- W4386072014 hasConcept C66322947 @default.
- W4386072014 hasConcept C86803240 @default.
- W4386072014 hasConceptScore W4386072014C108010975 @default.
- W4386072014 hasConceptScore W4386072014C113775141 @default.
- W4386072014 hasConceptScore W4386072014C119599485 @default.
- W4386072014 hasConceptScore W4386072014C119857082 @default.
- W4386072014 hasConceptScore W4386072014C127413603 @default.
- W4386072014 hasConceptScore W4386072014C154945302 @default.
- W4386072014 hasConceptScore W4386072014C165801399 @default.
- W4386072014 hasConceptScore W4386072014C41008148 @default.
- W4386072014 hasConceptScore W4386072014C6557445 @default.
- W4386072014 hasConceptScore W4386072014C66322947 @default.
- W4386072014 hasConceptScore W4386072014C86803240 @default.
- W4386072014 hasLocation W43860720141 @default.
- W4386072014 hasOpenAccess W4386072014 @default.
- W4386072014 hasPrimaryLocation W43860720141 @default.
- W4386072014 hasRelatedWork W2961085424 @default.
- W4386072014 hasRelatedWork W3046775127 @default.
- W4386072014 hasRelatedWork W3047144510 @default.
- W4386072014 hasRelatedWork W3170094116 @default.
- W4386072014 hasRelatedWork W4205958290 @default.
- W4386072014 hasRelatedWork W4285260836 @default.
- W4386072014 hasRelatedWork W4286629047 @default.
- W4386072014 hasRelatedWork W4306321456 @default.
- W4386072014 hasRelatedWork W4306674287 @default.
- W4386072014 hasRelatedWork W4224009465 @default.
- W4386072014 isParatext "false" @default.
- W4386072014 isRetracted "false" @default.
- W4386072014 workType "article" @default.