Research map / Label-efficient and robust learning / Self-supervised and contrastive
How Mask Matters: Towards Theoretical Understandings of Masked Autoencoders
Qi Zhang, Yifei Wang, Yisen Wang
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 changed | Where | Measured | Effect |
|---|---|---|---|
| add uniformity regularizer (U-MAE) | MAE ViT-Tiny · CIFAR-10 linear probing | accuracy (%): 59.6 -> 68.9 | Improved |
| add uniformity regularizer (U-MAE) | MAE ViT-Base · ImageNet-100 linear probing | accuracy (%): 61.2 -> 67.5 | Improved |
| add uniformity regularizer (U-MAE) | MAE ViT-Large · ImageNet-100 linear probing | accuracy (%): 64.4 -> 72.8 | Improved |
| add uniformity regularizer (U-MAE) | MAE ViT-Base · ImageNet-1K linear probing | accuracy (%): 55.4 -> 58.5 | Improved |
| add uniformity regularizer (U-MAE) | MAE ViT-Large · ImageNet-1K linear probing | accuracy (%): 62.2 -> 65.8 | Improved |
| add uniformity regularizer (U-MAE) | MAE · CIFAR-10 fine-tuning ViT-Tiny | accuracy (%): 89.6 -> 89.4 | No clear effect |
| add uniformity regularizer (U-MAE) | MAE ViT-Base · ImageNet-1K fine-tuning | accuracy (%): 82.9 -> 83.0 | No clear effect |
| add uniformity regularizer to SimMIM (U-SimMIM) | SimMIM ViT-Base · ImageNet-100 linear probing | accuracy (%): 54.3 -> 61.1 | Improved |
| set uniformity coefficient to 1e-3 (ablation) | U-MAE · ImageNet-100 linear probing | accuracy (%): 61.2 -> 65.9 | Improved |
| set uniformity coefficient to 1e-2 (ablation) | U-MAE · ImageNet-100 linear probing | accuracy (%): 61.2 -> 67.5 | Improved |
| set uniformity coefficient to 1e-1 (too large) (ablation) | U-MAE · ImageNet-100 linear probing | accuracy (%): 67.5 -> 45.1 | Hurt |
| set uniformity coefficient to 1e0 (ablation) | U-MAE · ImageNet-100 linear probing | accuracy (%): 67.5 -> 47.0 | Hurt |
Closest papers by meaning
- Learning Mask Invariant Mutual Information for Masked Image Modeling (ICLR 2025 · similarity 0.93)
- Self-Guided Masked Autoencoder (NeurIPS 2024 · similarity 0.92)
- Contrastive Tuning: A Little Help to Make Masked Autoencoders Forget (AAAI 2024 · similarity 0.91)
- The Dynamic Duo of Collaborative Masking and Target for Advanced Masked Autoencoder Learning (AAAI 2025 · similarity 0.91)
- R-MAE: Regions Meet Masked Autoencoders (ICLR 2024 · similarity 0.90)
- Extending Video Masked Autoencoders to 128 frames (NeurIPS 2024 · similarity 0.89)
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.
