Research map / Label-efficient and robust learning / Recognition and long-tail
Soft Labels for Ordinal Regression
Check 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 |
|---|---|---|---|
| encode ordinal ground-truth classes as metric-penalty soft labels rather than hard labels | VGG-16 for image-quality ranking · Image Aesthetics, Overall | Accuracy (%): 70.05 -> 72.03 | Helped |
| encode ordinal ground-truth classes as metric-penalty soft labels rather than hard labels | VGG-16 for image-quality ranking · Image Aesthetics, Overall | MAE: 0.316 -> 0.290 | Helped |
| encode age groups as metric-penalty soft labels rather than hard labels | VGG-16 for age estimation · Adience | Accuracy (%): 57.4 ± 5.8 -> 59.6 ± 3.6 | Helped |
| encode age groups as metric-penalty soft labels rather than hard labels | VGG-16 for age estimation · Adience | MAE: 0.55 ± 0.08 -> 0.49 ± 0.05 | Helped |
| use SORD for depth estimation | KITTI benchmark depth estimation · KITTI test | SILog: 11.77 -> 12.39 | Hurt |
| use SORD for multivariate horizon-parameter prediction | ResNet-50 horizon estimator · HLW | AUC (%): 82.33 -> 88.77 | Helped |
| predict with expected value instead of argmax (ablation) | SORD horizon estimator · HLW | AUC (%): 88.77 -> 89.98 | Helped |
| use squared log difference rather than squared difference as the depth penalty (ablation) | DeepLabv3+ SORD · KITTI validation | SILog: 10.32 -> 9.99 | Helped |
| use uniform depth discretization instead of SID (ablation) | DeepLabv3+ SORD · KITTI validation | SILog: 9.99 -> 10.68 | Hurt |
| use Cityscapes-pretrained weights instead of Pascal VOC-pretrained weights (ablation) | DeepLabv3+ SORD with squared log difference · KITTI validation | SILog: 9.99 -> 9.73 | Helped |
| increase SID intervals across the tested range (ablation) | SORD depth estimation · KITTI validation | SILog | Mixed |
| use expected-value prediction instead of argmax with the squared-difference soft labels (ablation) | DeepLabv3+ SORD depth estimator · KITTI validation | SILog: 10.32 -> 10.12 | Helped |
Closest papers by meaning
- A Constrained Deep Neural Network for Ordinal Regression (CVPR 2018 · similarity 0.91)
- Leaf-Smoothed Hierarchical Softmax for Ordinal Prediction (AAAI 2018 · similarity 0.89)
- SLACE: A Monotone and Balance-Sensitive Loss Function for Ordinal Regression (AAAI 2025 · similarity 0.88)
- Neural Collapse in Cumulative Link Models for Ordinal Regression: An Analysis with Unconstrained Feature Model (NeurIPS 2025 · similarity 0.86)
- CLOC: Contrastive Learning for Ordinal Classification with Multi-Margin N-pair Loss (CVPR 2025 · similarity 0.85)
- Leveraging Group Classification with Descending Soft Labeling for Deep Imbalanced Regression (AAAI 2025 · 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.