Research map / Vision-language and multimodal / Cross-modal learning
Deep Graph-neighbor Coherence Preserving Network for Unsupervised Cross-modal Hashing
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 |
|---|---|---|---|
| use DGCPN instead of the strongest listed baseline DSAH | cross-modal hashing · MIRFlickr-25K, 64-bit I2T | MAP: 0.722 -> 0.751 | Helped |
| use DGCPN instead of the strongest listed baseline DSAH | cross-modal hashing · MIRFlickr-25K, 64-bit T2I | MAP: 0.728 -> 0.749 | Helped |
| add graph-neighbor coherence to pairwise-distance similarity (ablation) | DGCPN · MIRFlickr-25K and NUS-WIDE, I2T and T2I | MAP: 0.725 -> 0.751; 0.724 -> 0.749; 0.626 -> 0.658; 0.632 -> 0.663 | Helped |
| add pairwise distances to graph-neighbor coherence (ablation) | DGCPN · MIRFlickr-25K and NUS-WIDE, I2T and T2I | MAP: 0.748 -> 0.751; 0.748 -> 0.749; 0.651 -> 0.658; 0.656 -> 0.663 | Helped |
| add coexistent-similarity and intra-/inter-modality consistency losses to graph-coherence loss alone (ablation) | DGCPN · MIRFlickr-25K and NUS-WIDE, I2T and T2I | MAP: 0.734 -> 0.751; 0.727 -> 0.749; 0.631 -> 0.658; 0.636 -> 0.663 | Helped |
| add intra- and inter-modality consistency loss (ablation) | DGCPN · MIRFlickr-25K and NUS-WIDE, I2T and T2I | MAP: 0.746 -> 0.751; 0.749 -> 0.749; 0.649 -> 0.658; 0.660 -> 0.663 | Helped |
| add half-real and half-binary optimization (ablation) | DGCPN · MIRFlickr-25K and NUS-WIDE, I2T and T2I | MAP: 0.746 -> 0.751; 0.731 -> 0.749; 0.651 -> 0.658; 0.655 -> 0.663 | Helped |
| replace half-real and half-binary optimization with adjusted Tanh (ablation) | DGCPN-Nhash · MIRFlickr-25K and NUS-WIDE, I2T and T2I | MAP: 0.746 -> 0.746; 0.739 -> 0.749; 0.656 -> 0.658; 0.662 -> 0.663 | Mixed |
| replace half-real and half-binary optimization with the hashing loss ||H-B||F (ablation) | DGCPN-Nhash · MIRFlickr-25K and NUS-WIDE, I2T and T2I | MAP: 0.748 -> 0.751; 0.742 -> 0.749; 0.653 -> 0.658; 0.658 -> 0.663 | Mixed |
Closest papers by meaning
- Vision-guided Text Mining for Unsupervised Cross-modal Hashing with Community Similarity Quantization (AAAI 2025 · similarity 0.90)
- UGGRH: Unsupervised Generative Completion and Graph-attention Refinement for Incomplete Cross-modal Hashing (NeurIPS 2026 · similarity 0.89)
- Statistical Model-driven Similarity Hashing: Bridging Modalities for Efficient Unsupervised Retrieval (AAAI 2025 · similarity 0.89)
- Lightweight Contrastive Distilled Hashing for Online Cross-modal Retrieval (AAAI 2025 · similarity 0.88)
- An End-To-End Graph Attention Network Hashing for Cross-Modal Retrieval (NeurIPS 2024 · similarity 0.88)
- Dual Self-Paced Cross-Modal Hashing (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.