Research map / Vision-language and multimodal / Cross-modal learning
ENHash: Error Notebook-Guided Fine-Grained Learning for Unsupervised Cross-Modal Hashing
Hao Fu, Zebing Yao, Chuangchuang Tan, Guanghua Gu
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 unsupervised incremental clustering instead of ordinary hierarchical clustering (ablation) | ENHash · MS COCO 64-bit | MAP@ALL I→T: 0.621 -> 0.669 | Improved |
| add unsupervised incremental clustering instead of ordinary hierarchical clustering (ablation) | ENHash · MS COCO 64-bit | MAP@ALL T→I: 0.627 -> 0.667 | Improved |
| add multi-cluster boundary loss to Ldis+Latt (ablation) | ENHash · MS COCO 64-bit | MAP@ALL I→T: 0.658 -> 0.669 | Improved |
| add multi-cluster boundary loss to Ldis+Latt (ablation) | ENHash · MS COCO 64-bit | MAP@ALL T→I: 0.662 -> 0.667 | Improved |
| add cross-modal attraction loss to Ldis+Lmcb (ablation) | ENHash · MS COCO 64-bit | MAP@ALL I→T: 0.659 -> 0.669 | Improved |
| add cross-modal attraction loss to Ldis+Lmcb (ablation) | ENHash · MS COCO 64-bit | MAP@ALL T→I: 0.662 -> 0.667 | Improved |
| add triplet discrimination loss to Latt+Lmcb (ablation) | ENHash · MS COCO 64-bit | MAP@ALL I→T: 0.369 -> 0.669 | Improved |
| use multi-cluster boundary loss instead of attraction loss alongside Ldis (ablation) | ENHash · MS COCO 64-bit | MAP@ALL I→T: 0.658 -> 0.659 | No clear effect |
| ENHash versus UCCH on text-query retrieval | ENHash · NUS-WIDE 16-bit and 32-bit | MAP@ALL T→I: 0.701 -> 0.735 (16-bit); 0.724 -> 0.759 (32-bit) | Improved |
| ENHash versus UCCH on image-query retrieval | ENHash · MS COCO 16-bit | MAP@ALL I→T: 0.605 -> 0.635 | Improved |
| ENHash versus SACH | ENHash · MIRFlickr 128-bit | MAP@ALL I→T: 0.734 -> 0.763 | Improved |
| ENHash versus supervised fine-grained CMCM on image-query | ENHash · FG-XMedia (128-bit) | MAP@ALL I→T: 0.349 -> 0.610 | Improved |
Closest papers by meaning
- Dual Self-Paced Cross-Modal Hashing (AAAI 2024 · similarity 0.88)
- Robust Self-Paced Hashing for Cross-Modal Retrieval with Noisy Labels (AAAI 2025 · similarity 0.88)
- Lightweight Contrastive Distilled Hashing for Online Cross-modal Retrieval (AAAI 2025 · similarity 0.87)
- Vision-guided Text Mining for Unsupervised Cross-modal Hashing with Community Similarity Quantization (AAAI 2025 · similarity 0.87)
- SHE: Streaming-media Hashing Retrieval (ICML 2025 · similarity 0.87)
- Statistical Model-driven Similarity Hashing: Bridging Modalities for Efficient Unsupervised Retrieval (AAAI 2025 · 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.
