Research map / Graph learning / Knowledge and temporal graphs
Scene Graph Embeddings Using Relative Similarity Supervision
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 Ranking loss instead of Triplet loss with Probability sampling | Scene graph GCN · Visual Genome test, per-image | Kendall τ: 0.187 -> 0.391 | Helped |
| Use Ranking loss instead of InfoNCE with Probability sampling | Scene graph GCN · Visual Genome test, noisy scene graph (noise=12) | MRR: 0.145 -> 0.857 | Helped |
| Use Probability sampling instead of Random sampling (ablation) | Ranking loss scene graph GCN · Visual Genome test, noisy scene graph (noise=12) | R@5: 0.789 -> 0.906 | Helped |
| Use Probability sampling instead of Extreme sampling (ablation) | Ranking loss scene graph GCN · Visual Genome test, noisy scene graph (noise=12) | R@5: 0.758 -> 0.906 | Helped |
| Use Probability sampling instead of Reject sampling (ablation) | Ranking loss scene graph GCN · Visual Genome test, noisy scene graph (noise=12) | R@5: 0.793 -> 0.906 | Helped |
| Use Extreme sampling instead of Random sampling (ablation) | Triplet loss scene graph GCN · Visual Genome test, noisy scene graph (noise=12) | MRR: 0.676 -> 0.431 | Hurt |
| Use Extreme sampling instead of Random sampling (ablation) | InfoNCE scene graph GCN · Visual Genome test, noisy scene graph (noise=12) | MRR: 0.615 -> 0.005 | Hurt |
Closest papers by meaning
- SCENIR: Visual Semantic Clarity through Unsupervised Scene Graph Retrieval (ICML 2025 · similarity 0.90)
- Object-Centric Representation Learning for Enhanced 3D Semantic Scene Graph Prediction (NeurIPS 2025 · similarity 0.87)
- Exploiting Distribution Constraints for Scalable and Efficient Image Retrieval (ICLR 2025 · similarity 0.86)
- RA-SGG: Retrieval-Augmented Scene Graph Generation Framework via Multi-Prototype Learning (AAAI 2025 · similarity 0.86)
- Composing Object Relations and Attributes for Image-Text Matching (CVPR 2024 · similarity 0.86)
- DSGG: Dense Relation Transformer for an End-to-end Scene Graph Generation (CVPR 2024 · similarity 0.86)
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.