AI Paper Atlas

Research map / Label-efficient and robust learning / Self-supervised and contrastive

Learning with Unmasked Tokens Drives Stronger Vision Learners

Taekyung Kim, Sanghyuk Chun, Byeongho Heo, Dongyoon Han

ECCV 2024 · Label-efficient and robust learning › Self-supervised and contrastive · approach: Masked autoencoder pretraining · 1 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 broader contextualization loss on unmasked tokens to MAEMAE ViT-B/16 · ImageNet-1K fine-tuningtop-1 accuracy: 83.7 -> 84.2Improved
add LUT to MAE at 400 epochs (ablation)MAE ViT-B/16 · ImageNet-1K ft (400 epochs)top-1 accuracy: 82.8 -> 83.5Improved
also steer masked tokens in addition to unmasked (ablation)LUT · ImageNet-1K ft (400 epochs)top-1 accuracy: 83.5 -> 83.0Hurt
use pooled visual tokens instead of cls-token as context (ablation)LUT · ImageNet-1K ft (400 epochs)top-1 accuracy: 83.2 -> 83.5Improved
add token-wise supervision on top of aggregated contextualization (ablation)LUT · ImageNet-1K ft (400 epochs)top-1 accuracy: 83.5 -> 83.1Hurt
use cosine distance instead of infoNCE or smoothed-l1 (ablation)LUT · ImageNet-1K ft (400 epochs)top-1 accuracy: 83.0 -> 83.5Improved
mask the target encoder input (ablation)LUT · ImageNet-1K ft (400 epochs)top-1 accuracy: 83.5 -> 82.5Hurt
use simple resized crop instead of random resized crop in LUT (ablation)LUT · ImageNet-1K ft (400 epochs)top-1 accuracy: 83.4 -> 83.5Improved
use simple resized crop instead of random resized crop in MAE (ablation)MAE · ImageNet-1K ft (400 epochs)top-1 accuracy: 82.8 -> 82.5Hurt
increase loss weight alpha from 0.25 to 1.0 (ablation)LUT · ImageNet-1K ft (400 epochs)top-1 accuracy: 83.5 -> 82.9Hurt
train with contextualization loss only, without MIM loss (ablation)LUT · ImageNet-1K ft (100 epochs)accuracy: 82.1 -> 82.0No clear effect
apply LUT to SimMIMSimMIM · ImageNet-1K ft (SimMIM, 100 epochs)accuracy: 81.6 -> 81.8Improved

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