Research map / Label-efficient and robust learning / Self-supervised and contrastive
Improving Transferability of Representations via Augmentation-Aware Self-Supervision
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 AugSelf auxiliary loss | SimSiam ResNet-18 (STL10) · Flowers | linear eval accuracy (%): 59.19 -> 72.18 | Helped |
| add AugSelf auxiliary loss | SimSiam ResNet-50 (ImageNet100) · Food | linear eval accuracy (%): 61.48 -> 65.63 | Helped |
| add AugSelf auxiliary loss | SimSiam ResNet-50 (ImageNet100) · Cars | linear eval accuracy (%): 48.20 -> 47.52 | Hurt |
| add AugSelf auxiliary loss | Supervised ResNet-50 · 11 linear-eval datasets | accuracy | Mixed |
| add AugSelf auxiliary loss | SimSiam ResNet-18 (STL10) · Plant Disease 5-way 1-shot | few-shot accuracy (%): 58.13 -> 62.85 | Helped |
| add AugSelf auxiliary loss | SimSiam · CUB200 bbox regression | l2 error: 0.00462 -> 0.00335 | Helped |
| add crop prediction to SimSiam (ablation) | SimSiam ResNet-18 · Flowers | linear eval accuracy (%): 59.19 -> 62.05 | Helped |
| add color prediction to SimSiam (ablation) | SimSiam ResNet-18 · Flowers | linear eval accuracy (%): 59.19 -> 71.08 | Helped |
| add crop and color prediction together vs color only (ablation) | SimSiam ResNet-18 · Flowers | linear eval accuracy (%): 71.08 -> 72.18 | Helped |
| pretrain with flip or blur prediction only (ablation) | ResNet-18 · Food | linear eval accuracy (%): 11.54 -> 9.69 (flip) | Hurt |
| add AugSelf to SimCLR/BYOL/SwAV (ablation) | ResNet-18 STL10 · Flowers | linear eval accuracy (%): 66.89 -> 74.07 (BYOL) | Helped |
| add AugSelf under rotation augmentation (ablation) | SimSiam ResNet-18 · CIFAR100 | linear eval accuracy (%): 50.24 -> 57.27 | Helped |
Closest papers by meaning
- Time to augment self-supervised visual representation learning (ICLR 2023 · similarity 0.89)
- An Augmentation-Aware Theory for Self-Supervised Contrastive Learning (ICML 2025 · similarity 0.89)
- Understanding Augmentation-based Self-Supervised Representation Learning via RKHS Approximation and Regression (ICLR 2024 · similarity 0.87)
- Can Generative Models Improve Self-Supervised Representation Learning? (AAAI 2025 · similarity 0.87)
- Harnessing small projectors and multiple views for efficient vision pretraining (NeurIPS 2024 · similarity 0.86)
- Effective Data Augmentation With Diffusion Models (ICLR 2024 · 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.