Research map / Vision-language and multimodal / Visual question answering
Multimodal Graph Networks for Compositional Generalization in Visual Question Answering
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 multimodal graph network (GCN over parsed graphs) instead of CNN+LSTM | caption truth prediction · Caption Truth Prediction, Dataset B (unseen) | accuracy (%): 79.7 (0.7) -> 98.2 (1.1) | Helped |
| MGN-GCN vs baselines on seen combinations | caption truth prediction · Caption Truth Prediction, Dataset A (seen) | accuracy (%): 99.3 (0.5) -> 99.6 (0.3) | No clear effect |
| add REINFORCE end-to-end fine-tuning after supervised pre-training (ablation) | MGN-e2e · CLOSURE | mean accuracy: 72.18% -> 80.87% | Helped |
| add supervised pre-training before RL fine-tuning (ablation) | MGN-e2e · CLOSURE | mean accuracy: 65.25% -> 80.87% | Helped |
| MGN-e2e instead of NS-VQA | CLOSURE · CLOSURE | mean accuracy: 77.19% -> 80.87% | Helped |
Closest papers by meaning
- Multimodal Commonsense Knowledge Distillation for Visual Question Answering (Student Abstract) (AAAI 2025 · similarity 0.90)
- On the generalization capacity of neural networks during generic multimodal reasoning (ICLR 2024 · similarity 0.89)
- Visual Graph Arena: Evaluating Visual Conceptualization of Vision and Multimodal Large Language Models (ICML 2025 · similarity 0.88)
- Compositional Chain-of-Thought Prompting for Large Multimodal Models (CVPR 2024 · similarity 0.88)
- Core-to-Global Reasoning for Compositional Visual Question Answering (AAAI 2025 · similarity 0.88)
- GRAPHGPT-O: Synergistic Multimodal Comprehension and Generation on Graphs (CVPR 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.