Research map / Vision-language and multimodal / Visual question answering
Correlation Information Bottleneck: Towards Adapting Pretrained Multimodal Models for Robust Visual Question Answering
Jingjing Jiang, Ziyi Liu, Nanning Zheng
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 CIB objective when finetuning | LXMERT · VQA-Rephrasings | VQA-Score (PER): 70.41 -> 72.62 | Improved |
| add CIB objective when finetuning | ViLBERT · VQA-Rephrasings | VQA-Score (PER): 59.16 -> 62.28 | Improved |
| add CIB objective when finetuning | BEiT-3-B · IV-VQA | #flips (lower better): 9.08 -> 5.40 | Improved |
| add CIB objective when finetuning | LXMERT · VQA-CE counterexamples | VQA-Score: 53.61 -> 57.14 | Improved |
| add CIB to finetuning on standard VQA v2 | ALBEF · VQA v2 test-dev | VQA-Score: 74.54 -> 76.27 | Improved |
| add CIB to finetuning on standard VQA v2 | UNITER-B · VQA v2 test-dev | VQA-Score: 72.70 -> 72.11 | Hurt |
| add CIB to finetuning | UNITER-B · AdVQA test | VQA-Score: 36.20 -> 37.85 | Improved |
| add CIB to weakly-supervised grounding | ALBEF-itm · RefCOCO+ val | accuracy: 58.46 -> 59.41 | Improved |
| add I(Xv;Tv)+I(Xl;Tl) compression with -I(Tv;Tl) and KL (full bound) vs only I(Y;T) (ablation) | LXMERT · VQA-Rephrasings | VQA-Score (PER): 70.41 -> 72.62 | Improved |
| add -I(Tv;Tl) modality-correlation term to the bound (ablation) | LXMERT · VQA-Rephrasings | VQA-Score (PER): 72.28 -> 72.62 | Improved |
| use CLUB instead of L1Out as MI upper-bound estimator (ablation) | LXMERT · VQA-Rephrasings | VQA-Score (PER): 72.31 -> 72.62 | Improved |
| use NWJ instead of InfoNCE as MI lower-bound estimator (ablation) | LXMERT · VQA-Rephrasings | VQA-Score (PER): 72.34 -> 72.62 | No clear effect |
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.
- BEiT-3-B — BEiT: BERT Pre-Training of Image Transformers (ICLR 2022) · better on 4 of 4 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
- Visual Instruction Bottleneck Tuning (NeurIPS 2025 · similarity 0.91)
- FRAMES-VQA: Benchmarking Fine-Tuning Robustness across Multi-Modal Shifts in Visual Question Answering (CVPR 2025 · similarity 0.91)
- Cross-modal Information Flow in Multimodal Large Language Models (CVPR 2025 · similarity 0.90)
- Consistency and Uncertainty: Identifying Unreliable Responses From Black-Box Vision-Language Models for Selective Visual Question Answering (CVPR 2024 · similarity 0.90)
- Rethinking Token Reduction for Large Vision-Language Models (CVPR 2026 · similarity 0.90)
- Predictive Regularization Against Visual Representation Degradation in Multimodal Large Language Models (CVPR 2026 · 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.
