Research map / Vision-language and multimodal / Visual question answering
CC-VQA: Conflict- and Correlation-Aware Method for Mitigating Knowledge Conflict in Knowledge-Based Visual Question Answering
Yuyang Hong, Jiaqi Gu, Yujing Lou, Lubin Fan, Qi Yang, Ying Wang, Kun Ding, Yue Wu, Shiming Xiang, Jieping Ye
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 Visual-Centric Contextual Conflict Reasoning (ablation) | CC-VQA · InfoSeek 10K subset (ablation) | VQA accuracy: 41.4 -> 43.3 | Improved |
| add Correlation-Enhanced Adaptive Decoding (ablation) | CC-VQA · InfoSeek 10K subset (ablation) | VQA accuracy: 43.3 -> 44.1 | Improved |
| add Correlation-Aware Positional Encoding (ablation) | CC-VQA · InfoSeek 10K subset (ablation) | VQA accuracy: 44.1 -> 45.0 | Improved |
| set position compression alpha to 0.5 instead of 1.0 (no compression) (ablation) | CC-VQA · InfoSeek 10K subset (ablation) | VQA accuracy: 44.1 -> 45.0 | Improved |
| compress low-relevance sentences more aggressively (alpha 0.1) (ablation) | CC-VQA · InfoSeek 10K subset (ablation) | VQA accuracy: 45.0 -> 44.0 | Hurt |
| apply CC-VQA conflict mitigation over vanilla RAG | Qwen2.5-VL-7B · InfoSeek 10K subset | harmful ratio (%): 10.53 -> 7.69 | Improved |
| apply CC-VQA conflict mitigation over vanilla RAG | Qwen2.5-VL-7B · InfoSeek 10K subset | helpful ratio (%): 16.82 -> 18.63 | Improved |
| apply CC-VQA with oracle articles | Qwen2.5-VL-7B · InfoSeek 10K subset, oracle articles | VQA accuracy: 55.3 -> 66.5 | Improved |
| apply CC-VQA over vanilla RAG | Qwen2.5-VL-7B · E-VQA single-hop | VQA accuracy: 36.7 -> 41.4 | Improved |
| apply CC-VQA over vanilla RAG | Qwen2.5-VL-7B · InfoSeek all | VQA accuracy: 41.8 -> 45.1 | Improved |
Later papers that compared with this method
1 later accepted paper reports CC-VQA as a baseline in their results tables; 1 of them report a better number on most of the results they share with it.
- ProMSA:Progressive Multimodal Search Agents for Knowledge-Based Visual Question Answering (ECCV 2026 · better on 2 of 2 shared results)
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
- Identifying and Resolving Pitfalls of Knowledge-Based VQA Benchmarks: Auditing, Repairing, and Augmenting (ECCV 2026 · similarity 0.93)
- REAL: Resolving Knowledge Conflicts in Knowledge-Intensive Visual Question Answering via Reasoning-Pivot Alignment (ICML 2026 · similarity 0.93)
- OK-VQA: A Visual Question Answering Benchmark Requiring External Knowledge (CVPR 2019 · similarity 0.92)
- Prompting Large Language Models With Answer Heuristics for Knowledge-Based Visual Question Answering (CVPR 2023 · similarity 0.92)
- Separation of Powers: On Segregating Knowledge from Observation in LLM-enabled Knowledge-based Visual Question Answering (CVPR 2025 · similarity 0.91)
- When Open-Vocabulary Visual Question Answering Meets Causal Adapter: Benchmark and Approach (AAAI 2025 · similarity 0.90)
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.
