Research map / Label-efficient and robust learning / Recognition and long-tail
Rank-N-Contrast: Learning Continuous Representations for Regression
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 |
|---|---|---|---|
| Learn the representation with RnC before L1 prediction instead of end-to-end L1 training | ResNet-18 regression · AgeDB | MAE: 6.63 -> 6.14 | Helped |
| Learn the representation with RnC before L1 prediction instead of end-to-end L1 training | 24-layer 1D ResNet regression · TUAB | MAE: 7.46 -> 6.97 | Helped |
| Learn the representation with RnC before L1 prediction instead of end-to-end L1 training | ResNet-18 regression · MPIIFaceGaze | Angular Error: 5.97 -> 5.27 | Helped |
| Learn the representation with RnC before L1 prediction instead of end-to-end L1 training | ResNet-18 regression · SkyFinder | MAE: 2.95 -> 2.86 | Helped |
| Use all 511 samples as positives rather than K=128 closest samples (ablation) | RnC · AgeDB | MAE: 6.46 -> 6.14 | Helped |
| Use all 511 samples as positives rather than K=256 closest samples (ablation) | RnC · AgeDB | MAE: 6.43 -> 6.14 | Helped |
| Use all 511 samples as positives rather than K=384 closest samples (ablation) | RnC · AgeDB | MAE: 6.29 -> 6.14 | Helped |
| Use negative L2 norm instead of cosine similarity (ablation) | RnC · AgeDB | MAE: 6.51 -> 6.14 | Helped |
| Use negative L2 norm instead of negative L1 norm (ablation) | RnC · AgeDB | MAE: 6.25 -> 6.14 | Helped |
| Freeze the encoder during predictor training rather than fine-tune it (ablation) | RnC · AgeDB | MAE: 6.36 -> 6.14 | Helped |
| Freeze the encoder during predictor training rather than jointly regularize (ablation) | RnC · AgeDB | MAE: 6.42 -> 6.14 | Helped |
Closest papers by meaning
- ACCon: Angle-Compensated Contrastive Regularizer for Deep Regression (AAAI 2025 · similarity 0.86)
- Contrastive Order Learning: A General Framework for Ordinal Regression (ICML 2026 · similarity 0.86)
- Improving Deep Regression with Tightness (ICLR 2025 · similarity 0.85)
- ConR: Contrastive Regularizer for Deep Imbalanced Regression (ICLR 2024 · similarity 0.84)
- Deep Regression Representation Learning with Topology (ICML 2024 · similarity 0.84)
- RankUp: Boosting Semi-Supervised Regression with an Auxiliary Ranking Classifier (NeurIPS 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.