Research map / Label-efficient and robust learning / Recognition and long-tail
HIER: Metric Learning Beyond Class Labels via Hierarchical Regularization
Sungyeon Kim, Boseung Jeong, Suha Kwak
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 HIER to PA in hyperbolic space (ablation) | Proxy Anchor · CUB, Cars, SOP, In-Shop | R@1: 74.7 -> 75.2; 84.3 -> 85.1; 82.3 -> 82.5; 90.4 -> 91.0 | Improved |
| Add HIER in spherical space to PA (ablation) | Proxy Anchor · CUB, Cars, SOP, In-Shop | R@1: 74.7 -> 75.1; 84.3 -> 84.8; 82.3 -> 82.4; 90.4 -> 90.6 | Improved |
| Use hyperbolic rather than spherical space for HIER with PA (ablation) | Proxy Anchor + HIER · CUB, Cars, SOP, In-Shop | R@1: 75.1 -> 75.2; 84.8 -> 85.1; 82.4 -> 82.5; 90.6 -> 91.0 | Improved |
| Add HIER in hyperbolic space to MS (ablation) | Multi Similarity · CUB, Cars, SOP, In-Shop | R@1: 75.4 -> 75.8; 83.5 -> 84.3; 80.0 -> 80.0; 88.1 -> 88.2 | Mixed |
| Add HIER in spherical space to MS (ablation) | Multi Similarity · CUB, Cars, SOP, In-Shop | R@1: 75.4 -> 75.6; 83.5 -> 83.6; 80.0 -> 80.0; 88.1 -> 88.0 | Mixed |
| Replace PA with MS while retaining HIER (ablation) | HIER · CUB, Cars, SOP, In-Shop | R@1: 75.2 -> 75.8; 85.1 -> 84.3; 82.5 -> 80.0; 91.0 -> 88.2 | Mixed |
| Use HIER with DeiT-S, 128 dimensions | Hyp · CUB, Cars, SOP, In-Shop | R@1: 74.7 -> 75.2; 82.1 -> 85.1; 83.0 -> 82.5; 90.9 -> 91.0 | Mixed |
Later papers that compared with this method
3 later accepted papers report HIER as a baseline in their results tables; 3 of them report a better number on most of the results they share with it.
- Potential Field Based Deep Metric Learning (CVPR 2025 · better on 6 of 6 shared results)
- Robust Hyperbolic Learning with Curvature-Aware Optimization (NeurIPS 2025 · better on 6 of 6 shared results)
- KAES: Multi-aspect Shared Knowledge Finding and Aligning for Cross-prompt Automated Scoring of Essay Traits (AAAI 2025 · better on 3 of 4 shared results)
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Hier-COS: Making Deep Features Hierarchy-aware via Composition of Orthogonal Subspaces (CVPR 2026 · similarity 0.88)
- No Fuss Distance Metric Learning Using Proxies (ICCV 2017 · similarity 0.87)
- Learning Structured Representations with Hyperbolic Embeddings (NeurIPS 2024 · similarity 0.87)
- Flattening the Parent Bias: Hierarchical Semantic Segmentation in the Poincare Ball (CVPR 2024 · similarity 0.85)
- Deep Disentangled Metric Learning (AAAI 2025 · similarity 0.85)
- Language Models as Hierarchy Encoders (NeurIPS 2024 · 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.
