Research map / Vision-language and multimodal / Multimodal LLMs
MissRAG: Addressing the Missing Modality Challenge in Multimodal Large Language Models
Vittorio Pipoli, Alessia Saporita, Federico Bolelli, Marcella Cornia, Lorenzo Baraldi, Costantino Grana, Rita Cucchiara, Elisa Ficarra
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 MissRAG+PE | ChatBridge · MUSIC-AVQA (missing video) | accuracy: 27.25 -> 43.18 | Improved |
| add MissRAG+PE | OneLLM · VALOR-32K (missing video) | CIDEr: 3.34 -> 16.57 | Improved |
| add MissRAG+PE | VideoLLaMA 2 · MUSIC-AVQA (missing video) | accuracy: 59.57 -> 60.11 | Improved |
| add MissRAG+PE | OneLLM · MOSI (missing audio) | accuracy: 48.54 -> 78.57 | Improved |
| add MissRAG+PE | VideoLLaMA 2 · MOSI (missing audio) | accuracy: 46.06 -> 33.38 | Hurt |
| add MissRAG+PE | VideoLLaMA 2 · MOSEI (missing audio) | accuracy: 41.36 -> 38.63 | Hurt |
| add prompt engineering alone (ablation) | OneLLM · MOSEI (missing video) | accuracy: 57.70 -> 64.58 | Improved |
| add MissRAG retrieval on top of PE (ablation) | OneLLM · MUSIC-AVQA (missing video) | accuracy: 41.81 -> 46.66 | Improved |
| add MissRAG retrieval on top of PE (ablation) | OneLLM · MOSI (missing video) | accuracy: 79.74 -> 74.49 | Hurt |
| add PE on top of MissRAG retrieval (ablation) | OneLLM · MUSIC-AVQA (missing video) | accuracy: 45.55 -> 46.66 | Improved |
| increase retrieved prototypes k from 1 to 5 (with PE) (ablation) | OneLLM · MUSIC-AVQA (missing video) | accuracy: 46.66 -> 47.29 | No clear effect |
| use merged prototype pool across datasets (ablation) | OneLLM · MUSIC-AVQA (missing video) | accuracy: 46.66 -> 46.57 | No clear effect |
Closest papers by meaning
- Assessing Modality Bias in Video Question Answering Benchmarks with Multimodal Large Language Models (AAAI 2025 · similarity 0.90)
- Retrieval-Augmented Dynamic Prompt Tuning for Incomplete Multimodal Learning (AAAI 2025 · similarity 0.90)
- Can Large Language Models Help Multimodal Language Analysis? MMLA: A Comprehensive Benchmark (NeurIPS 2025 · similarity 0.89)
- SEED-Bench: Benchmarking Multimodal Large Language Models (CVPR 2024 · similarity 0.89)
- SAE-V: Interpreting Multimodal Models for Enhanced Alignment (ICML 2025 · similarity 0.89)
- Bridging Modalities: Improving Universal Multimodal Retrieval by Multimodal Large Language Models (CVPR 2025 · similarity 0.89)
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.
