AI Paper Atlas

Research map / Vision-language and multimodal / Human and embodied modeling

SemMAE: Semantic-Guided Masking for Learning Masked Autoencoders

Gang Li, Heliang Zheng, Daqing Liu, Chaoyue Wang, Bing Su, Changwen Zheng

NeurIPS 2022 · Vision-language and multimodal › Human and embodied modeling · approach: Vision-language pretraining · 48 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 semantic-guided masking (learned parts) to MAEMAE ViT-Base 16x16 · ImageNet-1klinear probing accuracy: 63.7 -> 65.0Improved
add semantic-guided masking to MAE with 8x8 patchesMAE ViT-Base 8x8 · ImageNet-1klinear probing accuracy: 66.8 -> 68.7Improved
use 8x8 instead of 16x16 patch size (ablation)SemMAE · ImageNet-1kfine-tuning accuracy: 83.34 -> 84.50Improved
use 8x8 instead of 16x16 patch size (ablation)MAE · ImageNet-1kfine-tuning accuracy: 83.26 -> 83.10Hurt
use semantic parts from iBOT attention instead of baseline random masking (ablation)MAE · ImageNet-1klinear probing accuracy: 63.7 -> 63.6No clear effect
use self-supervised refined part learning instead of iBOT-initialized parts (ablation)SemMAE · ImageNet-1klinear probing accuracy: 63.6 -> 65.0Improved
mask 75% of whole parts instead of random masking (ablation)MAE · ImageNet-1klinear probing accuracy: 66.8 -> 52.9Hurt
mask 75% patches in each part instead of random masking (ablation)MAE · ImageNet-1klinear probing accuracy: 66.8 -> 66.5No clear effect
easy-to-hard adaptive masking (patches per part to whole parts) instead of random masking (ablation)MAE · ImageNet-1klinear probing accuracy: 66.8 -> 68.7Improved
reverse schedule (hard to easy, parts first) instead of easy to hard (ablation)SemMAE · ImageNet-1klinear probing accuracy: 68.7 -> 63.3Hurt
gamma 2 instead of 1/3 in masking schedule (ablation)SemMAE · ImageNet-1klinear probing accuracy: 66.2 -> 68.7Improved
gamma 3 instead of gamma 2 (ablation)SemMAE · ImageNet-1klinear probing accuracy: 68.7 -> 68.6No clear effect

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

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