Research map / Graph learning / Knowledge and temporal graphs
Reasoning Visual Dialogs With Structural and Partial Observations
Check 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 inferred dialog structure and iterative message passing rather than no EM iterations (ablation) | Visual dialog GNN · VisDial v0.9 | Recall@1: 46.73 -> 48.95 | Helped |
| Learn dynamic edge weights instead of fixing all edge weights to 1 (ablation) | Visual dialog GNN · VisDial v0.9 | Recall@1: 47.91 -> 48.95 | Helped |
| Use 3 EM iterations instead of 2 (ablation) | Visual dialog GNN · VisDial v0.9 | Recall@1: 48.18 -> 48.95 | Helped |
| Use 3 EM iterations instead of 4 (ablation) | Visual dialog GNN · VisDial v0.9 | Recall@1: 48.41 -> 48.95 | Helped |
| Use inferred dialog structure and iterative message passing rather than no EM iterations (ablation) | Visual dialog GNN · VisDial-Q | Recall@1: 25.69 -> 27.15 | Helped |
| Learn dynamic edge weights instead of fixing all edge weights to 1 (ablation) | Visual dialog GNN · VisDial-Q | Recall@1: 26.08 -> 27.15 | Helped |
| Use the proposed graph model instead of SCA | Visual dialog response ranking · VisDial v0.9 val | Recall@1: 50.29 -> 48.95 | Hurt |
| Use the proposed graph model instead of LF-Att | Visual dialog response ranking · VisDial v1.0 test-standard | MRR: 0.5707 -> 0.6137 | Helped |
| Use the proposed graph model instead of SF-QIH | Follow-up question ranking · VisDial-Q | Recall@1: 26.76 -> 27.15 | Helped |
Closest papers by meaning
- Iterative Context-Aware Graph Inference for Visual Dialog (CVPR 2020 · similarity 0.91)
- Factor Graph Attention (CVPR 2019 · similarity 0.89)
- Visual Dialog (CVPR 2017 · similarity 0.89)
- Image-Question-Answer Synergistic Network for Visual Dialog (CVPR 2019 · similarity 0.89)
- Structure-Aware Multimodal Sequential Learning for Visual Dialog (AAAI 2024 · similarity 0.87)
- Visual Graph Arena: Evaluating Visual Conceptualization of Vision and Multimodal Large Language Models (ICML 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.