Research map / Label-efficient and robust learning / Recognition and long-tail
Joining datasets via data augmentation in the label space for neural networks
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 |
|---|---|---|---|
| Use label-space augmentation instead of a standard FFN classifier | EfficientNet-B4 · Oxford-IIIT Pet | accuracy (%): 93.84 -> 94.66 | Helped |
| Join datasets using label-space augmentation | EfficientNet-B4 · PetFusion | accuracy (%): 91.50 -> 94.95 | Helped |
| Use label-space augmentation instead of a standard FFN classifier | EfficientNet-B4 · 102 Category Flower | accuracy (%): 90.91 -> 92.80 | Helped |
| Join datasets using label-space augmentation | EfficientNet-B4 · FlowerFusion | accuracy (%): 91.01 -> 93.27 | Helped |
| Augment the label graph | BERT · Arxiv Original | F1: 75.5 -> 77.4 | Helped |
| Join datasets using label-space augmentation | BERT · Arxiv Fusion | F1: 76.9 -> 79.3 | Helped |
| Augment the label graph | LSTM · Arxiv Original | F1: 69.9 -> 72.4 | Helped |
| Use a full rather than medium label graph, retaining the augmented nodes trimmed in the medium graph (ablation) | EfficientNet-B4 + label-aug · Oxford-IIIT Pet | accuracy difference from best result (%): -0.14 | Helped |
| Use a full rather than small label graph, retaining the augmented nodes trimmed in the small graph (ablation) | EfficientNet-B4 + label-aug · Oxford-IIIT Pet | accuracy difference from best result (%): -0.19 | Helped |
| Use a full rather than small label graph, retaining the augmented nodes trimmed in the small graph (ablation) | EfficientNet-B4 + label-aug · PetFusion | accuracy difference from best result (%): -0.1 | Helped |
| Use mean rather than sum pooling over sampled target paths (ablation) | EfficientNet-B4 + label-aug · Oxford-IIIT Pet | accuracy difference from best result (%): -0.71 | Helped |
| Use mean pooling rather than a random target path (ablation) | EfficientNet-B4 + label-aug · Oxford-IIIT Pet | accuracy difference from best result (%): -0.36 | Helped |
Closest papers by meaning
- Automated Label Unification for Multi-Dataset Semantic Segmentation with GNNs (NeurIPS 2024 · similarity 0.87)
- Knowledge Amalgamation for Multi-Label Classification via Label Dependency Transfer (AAAI 2023 · similarity 0.86)
- Label-Efficient Dataset Pruning via Semi-Supervised Pseudo-Labeling (NeurIPS 2026 · similarity 0.86)
- Adversarial Auto-Augment with Label Preservation: A Representation Learning Principle Guided Approach (NeurIPS 2022 · similarity 0.86)
- Candidate Label Set Pruning: A Data-centric Perspective for Deep Partial-label Learning (ICLR 2024 · similarity 0.85)
- Learning with Structural Labels for Learning with Noisy Labels (CVPR 2024 · similarity 0.84)
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.