Research map / Learning theory and algorithms / Prediction and decision losses
Structured Learning of Binary Codes with Column Generation for Optimizing Ranking Measures
Guosheng Lin, Fayao Liu, Chunhua Shen, Jianxin Wu, Heng Tao Shen
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 |
|---|---|---|---|
| use NDCG loss instead of AUC loss (ablation) | StructHash · STL10 | NDCG@100: 0.374 -> 0.435 | Improved |
| use NDCG loss instead of AUC loss (ablation) | StructHash · ISOLET | NDCG@100: 0.839 -> 0.881 | Improved |
| use NDCG loss instead of AUC loss (ablation) | StructHash · CIFAR | NDCG@100: 0.259 -> 0.335 | Improved |
| use NDCG loss instead of AUC loss (ablation) | StructHash · ISOLET | mAP: 0.796 -> 0.836 | Improved |
| use NDCG loss instead of AUC loss (ablation) | StructHash · CIFAR | mAP: 0.300 -> 0.294 | Hurt |
| use StructHash NDCG instead of triplet-loss CGHash | StructHash · MNIST | NDCG@100: 0.867 -> 0.851 | Hurt |
| use StructHash NDCG instead of triplet-loss CGHash | StructHash · ISOLET | NDCG@100: 0.866 -> 0.881 | Improved |
| use StructHash NDCG instead of SPLH | StructHash · ISOLET | NDCG@100: 0.629 -> 0.881 | Improved |
| use StructHash NDCG instead of SPLH | StructHash · CIFAR | NDCG@100: 0.357 -> 0.335 | Hurt |
| AUC-loss StructHash vs triplet-loss CGHash | StructHash · USPS | NDCG@100: 0.893 -> 0.900 | No clear effect |
| add stage-wise training (ablation) | StructHash · five datasets | training time | Improved |
| use simplified NDCG loss instead of NDCG with stage-wise training (ablation) | StructHash · five datasets | training time | Improved |
Closest papers by meaning
- Maximum-Margin Hamming Hashing (ICCV 2019 · similarity 0.90)
- Ordinal Constrained Binary Code Learning for Nearest Neighbor Search (AAAI 2017 · similarity 0.89)
- Auto-Encoding Twin-Bottleneck Hashing (CVPR 2020 · similarity 0.89)
- Hashing with Binary Matrix Pursuit (ECCV 2018 · similarity 0.89)
- Learning Attribute-Aware Hash Codes for Fine-Grained Image Retrieval via Query Optimization (ICML 2025 · similarity 0.87)
- Pairwise-Label-Based Deep Incremental Hashing with Simultaneous Code Expansion (AAAI 2024 · similarity 0.87)
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.
