Research map / Vision-language and multimodal / Visual question answering
Dynamic Memory Networks for Visual and Textual 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 |
|---|---|---|---|
| replace the original input module with input fusion (ablation) | DMN · bAbI-10k | mean error: 11.8 -> 3.9 | Helped |
| replace the original input module with input fusion (ablation) | DMN · DAQUAR-ALL | accuracy: 27.54 -> 28.43 | Helped |
| replace soft attention with attention-based GRU (ablation) | DMN2 · bAbI-10k | mean error: 3.9 -> 3.3 | Helped |
| replace soft attention with attention-based GRU (ablation) | DMN2 · DAQUAR-ALL | accuracy: 28.43 -> 28.62 | Helped |
| use untied weights and a ReLU memory update (ablation) | DMN3 · bAbI-10k | mean error: 3.3 -> 2.8 | Helped |
| use untied weights and a ReLU memory update (ablation) | DMN3 · DAQUAR-ALL | accuracy: 28.62 -> 28.79 | Helped |
| replace E2E with DMN+ | bAbI-10k · bAbI-10k | mean error (%): 4.2 -> 2.8 | Helped |
| replace SAN with DMN+ | VQA · VQA test-dev All | accuracy: 58.7 -> 60.3 | Helped |
| replace SAN with DMN+ | VQA · VQA test-standard All | accuracy: 58.9 -> 60.4 | Helped |
| use untied weights and a ReLU memory update rather than the simpler memory update | DMN+ · bAbI-10k QA16 basic induction | error | Hurt |
| use patch-based visual reasoning | DMN+ · VQA Number questions | accuracy | Hurt |
Closest papers by meaning
- Ask Me Anything: Dynamic Memory Networks for Natural Language Processing (ICML 2016 · similarity 0.92)
- Visual Question Answering With Memory-Augmented Networks (CVPR 2018 · similarity 0.91)
- Multi-Level Attention Networks for Visual Question Answering (CVPR 2017 · similarity 0.90)
- Learning Visual Knowledge Memory Networks for Visual Question Answering (CVPR 2018 · similarity 0.89)
- Focal Visual-Text Attention for Visual Question Answering (CVPR 2018 · similarity 0.89)
- Not Another Text Benchmark: Putting the “Visual" Back in Visual Question Answering for Large Video Models (NeurIPS 2026 · 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.