Research map / Label-efficient and robust learning / Self-supervised and contrastive
Tailoring Self-Supervision for Supervised Learning
WonJun Moon, Ji-Hwan Kim, Jae-Pil Heo
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 global rotation as data augmentation (ablation) | Baseline · CIFAR-10 classification | accuracy: 95.01 -> 92.76 | Hurt |
| add global rotation as multi-task (ablation) | Baseline · CIFAR-10 classification | accuracy: 95.01 -> 93.38 | Hurt |
| rotate only a local patch (LoRot-I) instead of the whole image (ablation) | Rotation task · CIFAR-10 affinity score | affinity: 58.06 -> 93.78 | Improved |
| add LoRot-I | Cross Entropy · CIFAR-10 OOD detection (LSUN FIX) | AUROC: 84.3 -> 94.4 | Improved |
| add LoRot-I | CSI · CIFAR-10 OOD detection (LSUN FIX) | AUROC: 92.1 -> 95.7 | Improved |
| add LoRot-E | LDAM-DRW · Imbalanced CIFAR-100 (ratio 0.01) | accuracy: 42.04 -> 46.48 | Improved |
| add LoRot-E | Adversarial training · CIFAR-10 PGD 20-step | accuracy: 46.5 -> 52.8 | Improved |
| add LoRot-E | ResNet50 baseline · ImageNet | top-1 accuracy: 76.32 -> 77.72 | Improved |
| add LoRot to CutMix | CutMix · ImageNet | top-1 accuracy | No clear effect |
| add LoRot-I | Mixup · ImageNet | top-1 accuracy: 77.58 -> 78.36 | Improved |
| add global rotation as multi-task (ablation) | SupCLR · CIFAR-100 | accuracy: 76.52 -> 71.80 | Hurt |
| add LoRot-I | SupCLR · CIFAR-100 | accuracy: 76.52 -> 78.78 | Improved |
Closest papers by meaning
- Towards Efficient and Effective Self-Supervised Learning of Visual Representations (ECCV 2022 · similarity 0.88)
- Extending and Analyzing Self-Supervised Learning Across Domains (ECCV 2020 · similarity 0.88)
- Self-Supervision Can Be a Good Few-Shot Learner (ECCV 2022 · similarity 0.88)
- Self-Supervised Representation Learning by Rotation Feature Decoupling (CVPR 2019 · similarity 0.86)
- A Survey on Self-Supervised Learning: Algorithms, Applications, and Future Trends (TPAMI 2024 · similarity 0.86)
- Towards Understanding Self-Pretraining for Sequence Classification (NeurIPS 2026 · similarity 0.85)
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.
