Research map / Vision-language and multimodal / Visual question answering
Visual Question Reasoning on General Dependency Tree
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 the dependency-tree ACMN without dataset-specific layout supervision | VQA reasoning on CLEVR · CLEVR test | overall accuracy: 83.7 -> 89.31 | Helped |
| use the dependency-tree ACMN for relational reasoning | Sort-of-CLEVR · Sort-of-CLEVR | relational accuracy: 94.0 -> 96.20 | Helped |
| use the dependency-tree ACMN | VQAv2 · VQAv2 test-dev | All accuracy: 65.32 -> 63.81 | Hurt |
| add the residual connection to the composition module (ablation) | ACMN · Sort-of-CLEVR | relational accuracy: 93.50 -> 96.20 | Helped |
| replace residual composition with residual and dense connections over prior hidden representations (ablation) | ACMN · Sort-of-CLEVR | relational accuracy: 96.20 -> 91.10 | Hurt |
| replace adversarial attention with the Relocate attention module (ablation) | ACMN · Sort-of-CLEVR | relational accuracy: 96.20 -> 90.10 | Hurt |
| concatenate image features with input attention maps instead of using an adversarial mask (ablation) | ACMN · Sort-of-CLEVR | relational accuracy: 96.20 -> 91.15 | Hurt |
Closest papers by meaning
- MUREL: Multimodal Relational Reasoning for Visual Question Answering (CVPR 2019 · similarity 0.90)
- Deep Modular Co-Attention Networks for Visual Question Answering (CVPR 2019 · similarity 0.90)
- Core-to-Global Reasoning for Compositional Visual Question Answering (AAAI 2025 · similarity 0.90)
- Detection-Based Intermediate Supervision for Visual Question Answering (AAAI 2024 · similarity 0.89)
- NaturalBench: Evaluating Vision-Language Models on Natural Adversarial Samples (NeurIPS 2024 · similarity 0.88)
- Exploring the Effectiveness of Object-Centric Representations in Visual Question Answering: Comparative Insights with Foundation Models (ICLR 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.