Research map / Vision-language and multimodal / Visual question answering
Logical Implications for Visual Question Answering Consistency
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 |
|---|---|---|---|
| Add logical-implication consistency loss | BAN · Introspect | accuracy: 67.14$\pm$0.10 -> 67.36$\pm$0.19 | Helped |
| Add logical-implication consistency loss | BAN · Introspect | consistency: 69.45$\pm$0.17 -> 70.38$\pm$0.39 | Helped |
| Add logical-implication consistency loss | LXMERT · Introspect | accuracy: 75.10$\pm$0.10 -> 75.17$\pm$0.08 | Helped |
| Add logical-implication consistency loss | LXMERT · Introspect | consistency: 76.24$\pm$0.63 -> 78.75$\pm$0.21 | Helped |
| Add logical-implication consistency loss | MVQA · DME test | accuracy (all): 81.15$\pm$0.49 -> 83.59$\pm$0.69 | Helped |
| Add logical-implication consistency loss | MVQA · DME test | consistency: 89.95$\pm$3.20 -> 95.78$\pm$1.19 | Helped |
| Use SQuINT consistency method | MVQA · DME test | accuracy (all): 81.15$\pm$0.49 -> 80.58$\pm$0.78 | Hurt |
| Use CP-VQA consistency method | MVQA · DME test | accuracy (all): 81.15$\pm$0.49 -> 83.49$\pm$0.99 | Helped |
| Flip an answer randomly in detected inconsistent pairs at test time | LXMERT · Introspect | accuracy: 75.10$\pm$0.10 -> 69.67$\pm$1.24 | Hurt |
| Flip the second answer in detected inconsistent pairs at test time | LXMERT · Introspect | consistency: 76.24$\pm$0.63 -> 87.56$\pm$2.51 | Helped |
| Increase the consistency-loss weight $\lambda$ (ablation) | LXMERT and MVQA · Introspect and DME | accuracy | Mixed |
| Increase the consistency-loss weight $\lambda$ (ablation) | LXMERT and MVQA · Introspect and DME | consistency | Helped |
Closest papers by meaning
- Cycle-Consistency for Robust Visual Question Answering (CVPR 2019 · similarity 0.94)
- Roses Are Red, Violets Are Blue... but Should VQA Expect Them To? (CVPR 2021 · similarity 0.93)
- Making the v in VQA Matter: Elevating the Role of Image Understanding in Visual Question Answering (CVPR 2017 · similarity 0.93)
- Towards Causal VQA: Revealing and Reducing Spurious Correlations by Invariant and Covariant Semantic Editing (CVPR 2020 · similarity 0.93)
- When Open-Vocabulary Visual Question Answering Meets Causal Adapter: Benchmark and Approach (AAAI 2025 · similarity 0.92)
- FRAMES-VQA: Benchmarking Fine-Tuning Robustness across Multi-Modal Shifts in Visual Question Answering (CVPR 2025 · similarity 0.91)
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.