AI Paper Atlas

Research map / Architectures and efficiency / Attention and transformers

Scratching Visual Transformer's Back with Uniform Attention

Nam Hyeon-Woo, Kim Yu-Ji, Byeongho Heo, Dongyoon Han, Seong Joon Oh, Tae-Hyun Oh

ICCV 2023 · Architectures and efficiency › Attention and transformers · approach: Hybrid and interpretable ViT designs · 40 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add Context Broadcasting (CB) to ViT-SViT-S · ImageNet-1K valtop-1 accuracy: 79.9 -> 80.8Improved
insert CB at end of MLP block rather than in MSA or both (ablation)CB in ViT-S · ImageNet-1K valtop-1 accuracy: 80.1 -> 80.5Improved
place CB at Mid or End of MLP rather than Front (ablation)CB in ViT-S · ImageNet-1K valtop-1 accuracy: 79.9 -> 80.5Improved
insert CB only in upper layers instead of all layers (ablation)CB · ImageNet-1K val, ViT-Ti/S/Btop-1 accuracy: 80.5 -> 80.8 (ViT-S)Improved
compare CB against Squeeze-and-Excitation moduleViT-S · ImageNet-1K valtop-1 accuracy: 80.3 -> 80.8Improved
reduce number of attention heads with CB (ablation)ViT-Ti with CB · ImageNet-1K valaccuracy gapImproved
add CB to ViT-S for attention entropyViT-S · layer-wise attention entropyentropyImproved
add CB to ViT backbones for segmentationViT-Ti/S/B UperNet · ADE20K 40K iterationsmIoU: +1.0, +0.4, +0.8Improved
add CB_S to ViT-S for segmentationViT-S UperNet · ADE20K 160K iterationsmIoU: 43.3 -> 43.1Hurt
add CB to ViT-Ti for detectionViT-Ti Mask R-CNN · COCO valAP box: 34.8 -> 35.1Improved
add CB to ViT-S for robustnessViT-S · FGSM attackaccuracy: 27.2 -> 32.3Improved
add CB_S to ViLT image tokens, text tokens or bothViLT · VQAv2accuracy: 71.28 -> 71.46Improved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy