AI Paper Atlas

Research map / Architectures and efficiency / Attention and transformers

Adaptive Token Sampling for Efficient Vision Transformers

Mohsen Fayyaz, Soroush Abbasi Koohpayegani, Farnoush Rezaei Jafari, Sunando Sengupta, Hamid Reza Vaezi Joze, Eric Sommerlade, Hamed Pirsiavash, Jürgen Gall

ECCV 2022 · Architectures and efficiency › Attention and transformers · approach: Efficient and robust vision transformers · 223 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 ATS to DeiT-S (stages 3-11) with fine-tuningDeiT-S · ImageNet-1K valGFLOPs: 4.6 -> 2.9Improved
add ATS to DeiT-S (stages 3-11) with fine-tuningDeiT-S · ImageNet-1K valtop-1 accuracy (%): 79.8 -> 79.7No clear effect
add ATS to CvT-21CvT-21 · ImageNet-1K valGFLOPs: 7.1 -> 5.1Improved
add ATS to TimeSformer-LTimeSformer-L · Kinetics-400GFLOPs: 7,140 -> 3,510Improved
add ATS to X-ViT 16xX-ViT · Kinetics-400GFLOPs: 425 -> 259Improved
add ATS to EViT-DeiT-SEViT-DeiT-S · ImageNet-1K valGFLOPs: 3.0 -> 2.5Improved
use classification-token attention rather than summed self-attention as significance score (ablation)ATS DeiT-S · ImageNet-1K valtop-1 accuracy (%)Improved
use classification-token attention rather than a random token's attention (ablation)ATS DeiT-S · ImageNet-1K valtop-1 accuracy (%)Improved
add L2 norm of values to the significance score (ablation)ATS DeiT-S · ImageNet-1K valtop-1 accuracy (%): about 0.2%Improved
use inverse transform sampling instead of Top-K selection (ablation)ATS DeiT-S · ImageNet-1K valtop-1 accuracy (%)Improved
fine-tune DeiT-S+ATS (ablation)DeiT-S+ATS · ImageNet-1K valtop-1 accuracy (%)Improved
fine-tune XViT+ATS (ablation)XViT+ATS 16x · Kinetics-600top-1 accuracy (%): 83.4 -> 84.4Improved

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