Research map / Label-efficient and robust learning / Semi-supervised learning
Multi-Task Curriculum Framework for Open-Set Semi-Supervised Learning
Qing Yu, Daiki Ikami, Go Irie, Kiyoharu Aizawa
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 joint OOD-score optimization with curriculum selection | MixMatch · CIFAR-10, 250 labels, LSUN as OOD | accuracy (%): 76.32 -> 86.65 | Improved |
| add joint OOD-score optimization with curriculum selection | MixMatch · CIFAR-10, 250 labels, Uniform as OOD | accuracy (%): 72.90 -> 85.54 | Improved |
| add joint OOD-score optimization with curriculum selection | MixMatch · CIFAR-10, 250 labels, TIN as OOD | accuracy (%): 82.42 -> 86.44 | Improved |
| add joint OOD-score optimization with curriculum selection | MixMatch · CIFAR-10, 4000 labels, TIN as OOD | accuracy (%): 91.25 -> 93.03 | Improved |
| add joint OOD-score optimization with curriculum selection | MixMatch · SVHN, 250 labels, TIN as OOD | accuracy (%): 94.66 -> 95.21 | Improved |
| add method when only 2000 OOD samples are in unlabeled data | MixMatch · CIFAR-10, 250 labels, LSUN OOD, 2000 OOD samples | accuracy (%): 88.16 -> 84.82 | Hurt |
| increase number of OOD samples from 2000 to 20000 (baseline degrades, ours stable) (ablation) | MixMatch · CIFAR-10, 250 labels, LSUN OOD | accuracy (%): 88.16 -> 70.3 (baseline); 84.82 -> 85.83 (ours) | Mixed |
| replace softmax or ODIN detection with joint-optimized OOD score (ablation) | Hendrycks & Gimpel / ODIN · CIFAR-10 ID, TIN OOD (250 labels) | OOD detection AUROC (%): 54.54 -> 98.86 | Improved |
| replace softmax or ODIN detection with joint-optimized OOD score (ablation) | Hendrycks & Gimpel / ODIN · SVHN ID, Uniform OOD (250 labels) | OOD detection AUROC (%): 2.79 -> 99.97 | Improved |
| replace UASD confidence filtering with explicit OOD detector trained on unlabeled data (ablation) | UASD · CIFAR-10, 250 labels, LSUN as OOD | accuracy (%): 80.87 -> 86.65 | Improved |
| replace UASD confidence filtering with explicit OOD detector trained on unlabeled data (ablation) | UASD · CIFAR-10, 250 labels, TIN as OOD | detection recall (%): 66.47 -> 99.22 | Improved |
| use natural-image outliers (TIN/LSUN) instead of synthetic noise outliers (ablation) | MixMatch · CIFAR-10, 250 labels, baseline accuracy | accuracy (%): 72.90 -> 82.42 | Improved |
Closest papers by meaning
- Exploration and Exploitation of Unlabeled Data for Open-Set Semi-supervised Learning (IJCV 2024 · similarity 0.92)
- Trash To Treasure: Harvesting OOD Data With Cross-Modal Matching for Open-Set Semi-Supervised Learning (ICCV 2021 · similarity 0.90)
- STEP: Out-of-Distribution Detection in the Presence of Limited In-Distribution Labeled Data (NeurIPS 2021 · similarity 0.90)
- Rethinking Safe Semi-supervised Learning: Transferring the Open-set Problem to A Close-set One (ICCV 2023 · similarity 0.89)
- Robust Semi-Supervised Learning by Wisely Leveraging Open-Set Data (TPAMI 2024 · similarity 0.89)
- Targeted Representation Alignment for Open-World Semi-Supervised Learning (CVPR 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.
