AI Paper Atlas

Research map / Architectures and efficiency / Attention and transformers

KVT: k-NN Attention for Boosting Vision Transformers

Pichao Wang, Xue Wang, Fan Wang, Ming Lin, Shuning Chang, Hao Li, Rong Jin

ECCV 2022 · Architectures and efficiency › Attention and transformers · approach: Efficient and robust vision transformers · 151 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
replace dense attention with k-NN attentionDeiT-Tiny · ImageNet-1K valtop-1: 72.2% -> 73.0%Improved
replace dense attention with k-NN attentionPiT-Base · ImageNet-1K valtop-1: 82.0% -> 82.6%Improved
replace dense attention with k-NN attentionSwin-Tiny · ImageNet-1K valtop-1: 81.2% -> 81.3%No clear effect
replace dense attention with k-NN attentionSwin-T · COCO valmAP(box): 46.0 -> 46.2Improved
replace dense attention with k-NN attentionTwins-SVT-Base · ADE20K valmIoU: 47.4 -> 47.9Improved
use k-NN attention with k about half of the tokens instead of dense attention (ablation)DeiT-Tiny · ImageNet-1K valtop-1: 72.2% -> 73.0%Improved
add k-NN attention for faster early convergence at 10 epochs (ablation)DeiT-S · ImageNet-1K valtop-1: 29.1% -> 31.3%Improved
add k-NN attention at 10 epochs (ablation)CvT-13 · ImageNet-1K valtop-1: 51.4% -> 54.2%Improved
add k-NN attention at 300 epochs (ablation)DeiT-S · ImageNet-1K valtop-1: 79.8% -> 80.0%Improved
use k-NN attention instead of softmax temperature (ablation)DeiT-Tiny · ImageNet-1K valtop-1: 72.5 -> 73.0Improved
use small softmax temperature t=0.05 or 0.1 (ablation)DeiT-Tiny softmax · ImageNet-1K valtop-1: 72.2 -> crashHurt
use larger softmax temperature t=0.75 (ablation)DeiT-Tiny softmax · ImageNet-1K valtop-1: 72.2 -> 72.5No clear effect

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