Research map / Detection and segmentation / CNN features
MST: Masked Self-Supervised Transformer for Visual Representation
Zhaowen Li, Zhiyang Chen, Fan Yang, Wei Li, Yousong Zhu, Chaoyang Zhao, Rui Deng, Liwei Wu, Rui Zhao, Ming Tang, Jinqiao 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 random token masking (ablation) | MST (DeiT-S, 100 epochs) · ImageNet linear probe | top-1 accuracy (%): 73.1 -> 63.2 | Hurt |
| add attention-guided token masking (ablation) | MST (DeiT-S, 100 epochs) · ImageNet linear probe | top-1 accuracy (%): 73.1 -> 73.7 | Improved |
| choose mask candidates from lowest 1/8 attention tokens instead of all tokens (ablation) | MST attention-guided mask · ImageNet linear probe | top-1 accuracy (%): 63.2 -> 73.6 | Improved |
| use BN in projection head with MST update rule (ablation) | MST · ImageNet linear probe | top-1 accuracy (%): 73.1 -> 73.9 | Improved |
| use BN in projection head (ablation) | baseline (no MST) · ImageNet linear probe | top-1 accuracy (%): 72.4 -> 71.6 | Hurt |
| add global image decoder with original-image reconstruction (vs masked tokens only) with attention mask (ablation) | MST · ImageNet linear probe | top-1 accuracy (%): 71.9 -> 73.5 | Improved |
| use MST mask replacement instead of BERT-style replacement (ablation) | MST · ImageNet linear probe | top-1 accuracy (%): 73.5 -> 73.9 | Improved |
| vary restoration loss weight lambda2 (0.2 to 0.8) (ablation) | MST · ImageNet linear probe | top-1 accuracy (%): 73.7, 73.5, 73.9, 73.6 | No clear effect |
| train 300 epochs instead of 100 | MST (DeiT-S) · ImageNet linear probe | top-1 accuracy (%): 75.0 -> 76.9 | Improved |
| use k=200 instead of k=10 in k-NN (ablation) | MST · ImageNet k-NN | top-1 accuracy (%): 75.0 -> 71.8 | Hurt |
Closest papers by meaning
- Good Helper Is around You: Attention-Driven Masked Image Modeling (AAAI 2023 · similarity 0.88)
- What to Hide from Your Students: Attention-Guided Masked Image Modeling (ECCV 2022 · similarity 0.88)
- Vision Transformers Need Registers (ICLR 2024 · similarity 0.87)
- Closest Neighbors are Harmful for Lightweight Masked Auto-encoders (CVPR 2025 · similarity 0.87)
- Video Language Model Pretraining with Spatio-temporal Masking (CVPR 2025 · similarity 0.87)
- A Theoretical Analysis of Self-Supervised Learning for Vision Transformers (ICLR 2025 · similarity 0.86)
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.
