Research map / Vision-language and multimodal / Visual question answering
Query and Attention Augmentation for Knowledge-Based Explainable Reasoning
Yifeng Zhang, Ming Jiang, Qi Zhao
Check your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (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 |
|---|---|---|---|
| Add query and attention augmentation | NSM · OK-VQA | accuracy (%): 26.79 -> 29.24 | Improved |
| Add query and attention augmentation | NSM · FVQA | accuracy (%): 64.08 -> 68.74 | Improved |
| Add query and attention augmentation | NSM · GQA | accuracy (%): 63.17 -> 63.82 | Improved |
| Add query and attention augmentation | NSM · VQA v2 | accuracy (%): 65.77 -> 67.69 | Improved |
| Add memory-augmented attention (ablation) | NSM · OK-VQA | accuracy (%): 26.79 -> 27.91 | Improved |
| Add visual-only knowledge-augmented queries (ablation) | NSM · OK-VQA | accuracy (%): 26.79 -> 28.23 | Improved |
| Add external-only knowledge-augmented queries (ablation) | NSM · OK-VQA | accuracy (%): 26.79 -> 27.86 | Improved |
| Add knowledge-augmented queries for both knowledge sources (ablation) | NSM · OK-VQA | accuracy (%): 26.79 -> 28.42 | Improved |
| Replace reinforcement-learning query augmentation with cosine-similarity query expansion (ablation) | NSM · OK-VQA | accuracy (%): 28.37 -> 29.24 | Hurt |
| Use both visual and external knowledge bases rather than one knowledge base alone (ablation) | NSM + Ours · OK-VQA | accuracy (%): 28.73; 28.59; 28.26 -> 29.24 | Improved |
Closest papers by meaning
- Knowledge Acquisition for Visual Question Answering via Iterative Querying (CVPR 2017 · similarity 0.91)
- Inverse Visual Question Answering: A New Benchmark and VQA Diagnosis Tool (TPAMI 2018 · similarity 0.90)
- How Transferable Are Reasoning Patterns in VQA? (CVPR 2021 · similarity 0.90)
- ReAG: Reasoning-Augmented Generation for Knowledge-based Visual Question Answering (CVPR 2026 · similarity 0.90)
- VQA-E: Explaining, Elaborating, and Enhancing Your Answers for Visual Questions (ECCV 2018 · similarity 0.89)
- Knowledge-based Visual Question Answer with Multimodal Processing, Retrieval and Filtering (NeurIPS 2025 · similarity 0.88)
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.
