Research map / Vision-language and multimodal / Visual question answering
Interpretable Visual Reasoning via Probabilistic Formulation under Natural Supervision
Xinzhe Han, Shuhui Wang, Chi Su, Weigang Zhang, Qingming Huang, Qi Tian
Read the paperCheck 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 TRN to UpDn | UpDn · VQA v2.0 test-dev | accuracy (All): 65.82 -> 67.00 | Improved |
| add TRN to BAN-3 | BAN-3 · VQA v2.0 test-dev | accuracy (All): 68.43 -> 68.38 | No clear effect |
| add TRN to BAN-3 | BAN-3 · VQA v2.0 test-std | accuracy: 68.47 -> 68.76 | Improved |
| add TRN to UpDn | UpDn · CLEVR val | accuracy: 78.1% -> 87.7% | Improved |
| add TRN to UpDn | UpDn · CLEVR-Humans | accuracy: 56.6% -> 69.5% | Improved |
| add TRN to BAN-5 | BAN-5 · CLEVR val | accuracy: 83.1% -> 85.2% | Improved |
| add TRN to BAN-5 | BAN-5 · CLEVR-Humans | accuracy: 60.5% -> 65.4% | Improved |
| add TRN to DFAF-5 | DFAF-5 · CLEVR val | accuracy: 95.5% -> 96.7% | Improved |
| add TRN to DFAF-5 | DFAF-5 · CLEVR-Humans | accuracy: 63.2% -> 72.9% | Improved |
| add Generative Reconstruction to UpDn+TRN (ablation) | UpDn+TRN · CLEVR val | accuracy: 82.3% -> 87.7% | Improved |
| add State Inference and State Transition to UpDn (ablation) | UpDn+TRN · CLEVR val | accuracy: 78.1% -> 87.7% | Improved |
| use a different partial component subset (Table 3 rows 84.3 and 77.6) versus full TRN (ablation) | UpDn+TRN · CLEVR val | accuracy: 84.3% -> 87.7% | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- MuRel — MUREL: Multimodal Relational Reasoning for Visual Question Answering (CVPR 2019) · better on 1 of 1 shared result
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- SLVR: Structured Latent Visual Reasoning via Human-like Reasoning Flows (NeurIPS 2026 · similarity 0.90)
- StaR-KVQA: Structured Reasoning Traces for Implicit-Knowledge Visual Question Answering (CVPR 2026 · similarity 0.90)
- Not Another Text Benchmark: Putting the “Visual" Back in Visual Question Answering for Large Video Models (NeurIPS 2026 · similarity 0.88)
- When Open-Vocabulary Visual Question Answering Meets Causal Adapter: Benchmark and Approach (AAAI 2025 · similarity 0.88)
- Latent Visual Reasoning (ICLR 2026 · similarity 0.88)
- Semantic-Enriched Latent Visual Reasoning (ICML 2026 · 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.
