Research map / Vision-language and multimodal / Visual question answering
Neural Module Networks
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 |
|---|---|---|---|
| compose reusable neural modules according to question structure | VIS+LSTM on Shapes All · Shapes All | accuracy: 65.3 -> 90.6 | Helped |
| compose reusable neural modules according to question structure | VIS+LSTM on Shapes size 6 · Shapes size 6 | accuracy: 61.7 -> 85.2 | Helped |
| train without size-6 questions (ablation) | NMN on Shapes · Shapes All | accuracy: 90.6 -> 90.8 | No clear effect |
| add the whole-question LSTM to NMN (ablation) | NMN on VQA test-dev · VQA test-dev All | accuracy: 42.7 -> 54.8 | Helped |
| add the whole-question LSTM to NMN (ablation) | NMN on VQA test-dev · VQA test-dev Yes/No | accuracy: 69.38 -> 77.7 | Helped |
| add the whole-question LSTM to NMN (ablation) | NMN on VQA test-dev · VQA test-dev Number | accuracy: 30.7 -> 37.2 | Helped |
| add the whole-question LSTM to NMN (ablation) | NMN on VQA test-dev · VQA test-dev Other | accuracy: 22.7 -> 39.3 | Helped |
| use NMN+LSTM instead of the LSTM baseline | VQA test-dev Yes/No · VQA test-dev Yes/No | accuracy: 78.20 -> 77.7 | Hurt |
Closest papers by meaning
- Graph-Structured Representations for Visual Question Answering (CVPR 2017 · similarity 0.90)
- Deep Modular Co-Attention Networks for Visual Question Answering (CVPR 2019 · similarity 0.90)
- Visual Question Reasoning on General Dependency Tree (CVPR 2018 · similarity 0.90)
- How Modular should Neural Module Networks Be for Systematic Generalization? (NeurIPS 2021 · similarity 0.89)
- Multimodal Graph Networks for Compositional Generalization in Visual Question Answering (NeurIPS 2020 · similarity 0.89)
- Ask Me Anything: Free-Form Visual Question Answering Based on Knowledge From External Sources (CVPR 2016 · similarity 0.89)
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.