# MiMIC: Mitigating Visual Modality Collapse in Universal Multimodal Retrieval While Avoiding Semantic Misalignment

arXiv: https://arxiv.org/abs/2604.21326 · HTML: https://mlsanity.com/paper/2604.21326/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 24 of its 33 references that are in the map.
- First posted on arXiv 23 Apr 2026.

## Builds on (most cited first, 24 of 24)

- [Learning Transferable Visual Models From Natural Language Supervision](https://mlsanity.com/paper/2103.00020/) (2021)
- [BART: Denoising Sequence-to-Sequence Pre-training for Natural Language Generation, Translation, and Comprehension](https://mlsanity.com/paper/1910.13461/) (2019)
- [Dense Passage Retrieval for Open-Domain Question Answering](https://mlsanity.com/paper/2004.04906/) (2020)
- [Sigmoid Loss for Language Image Pre-Training](https://mlsanity.com/paper/2303.15343/) (2023)
- [Qwen3 Embedding: Advancing Text Embedding and Reranking Through Foundation Models](https://mlsanity.com/paper/2506.05176/) (2025)
- [OK-VQA: A Visual Question Answering Benchmark Requiring External Knowledge](https://mlsanity.com/paper/1906.00067/) (2019)
- [Approximate Nearest Neighbor Negative Contrastive Learning for Dense Text Retrieval](https://mlsanity.com/paper/2007.00808/) (2020)
- [LLM2Vec: Large Language Models Are Secretly Powerful Text Encoders](https://mlsanity.com/paper/2404.05961/) (2024)
- [What Makes Training Multi-Modal Classification Networks Hard?](https://mlsanity.com/paper/1905.12681/) (2019)
- [PaLI-X: On Scaling up a Multilingual Vision and Language Model](https://mlsanity.com/paper/2305.18565/) (2023)
- [Can Pre-trained Vision and Language Models Answer Visual Information-Seeking Questions?](https://mlsanity.com/paper/2302.11713/) (2023)
- [Are Multimodal Transformers Robust to Missing Modality?](https://mlsanity.com/paper/2204.05454/) (2022)
- [GME: Improving Universal Multimodal Retrieval by Multimodal LLMs](https://mlsanity.com/paper/2412.16855/) (2024)
- [Encyclopedic VQA: Visual questions about detailed properties of fine-grained categories](https://mlsanity.com/paper/2306.09224/) (2023)
- [VISTA: Visualized Text Embedding For Universal Multi-Modal Retrieval](https://mlsanity.com/paper/2406.04292/) (2024)
- [Universal Vision-Language Dense Retrieval: Learning A Unified Representation Space for Multi-Modal Retrieval](https://mlsanity.com/paper/2209.00179/) (2022)
- [Mitigating Modality Collapse in Multimodal VAEs via Impartial Optimization](https://mlsanity.com/paper/2206.04496/) (2022)
- [Encoder-Decoder Gemma: Improving the Quality-Efficiency Trade-Off via Adaptation](https://mlsanity.com/paper/2504.06225/) (2025)
- [A Closer Look at Multimodal Representation Collapse](https://mlsanity.com/paper/2505.22483/) (2025)
- [End-to-end Knowledge Retrieval with Multi-modal Queries](https://mlsanity.com/paper/2306.00424/) (2023)
- [MARVEL: Unlocking the Multi-Modal Capability of Dense Retrieval via Visual Module Plugin](https://mlsanity.com/paper/2310.14037/) (2023)
- [When Language Overrules: Revealing Text Dominance in Multimodal Large Language Models](https://mlsanity.com/paper/2508.10552/) (2025)
- [Do We Really Need to Drop Items with Missing Modalities in Multimodal Recommendation?](https://mlsanity.com/paper/2408.11767/) (2024)
- [Learning to Compress: Unlocking the Potential of Large Language Models for Text Representation](https://mlsanity.com/paper/2511.17129/) (2025)

## Abstract (arXiv)

Universal Multimodal Retrieval (UMR) aims to map different modalities (e.g., visual and textual) into a shared embedding space for multi-modal retrieval. Existing UMR methods can be broadly divided into two categories: early-fusion approaches, such as Marvel, which projects visual features into the language model (LM) space for integrating with text modality, and late-fusion approaches, such as UniVL-DR, which encode visual and textual inputs using separate encoders and obtain fused embeddings through addition. Our pilot study reveals that Marvel exhibits visual modality collapse, which is characterized by the model's tendency to disregard visual features while depending excessively on textual cues. In contrast, although UniVL-DR is less affected by this issue, it is more susceptible to semantic misalignment, where semantically related content is positioned far apart in the embedding space. To address these challenges, we propose MiMIC, which introduces two key innovations: (1) a fusion-in-decoder architecture for effective multimodal integration, and (2) robust training through single modality mixin and random caption dropout. Experiments on the WebQA+ and EVQA+ datasets, where image in documents or queries might lack captions, indicate that MiMIC consistently outperforms both early- and late-fusion baselines.