@inproceedings{huang-etal-2025-enhancing-multimodal,
title = "Enhancing Multimodal Unified Representations for Cross Modal Generalization",
author = "Huang, Hai and
Xia, Yan and
Ji, Shengpeng and
Wang, Shulei and
Wang, Hanting and
Fang, Minghui and
Zhu, Jieming and
Dong, Zhenhua and
Zhou, Sashuai and
Zhao, Zhou",
editor = "Che, Wanxiang and
Nabende, Joyce and
Shutova, Ekaterina and
Pilehvar, Mohammad Taher",
booktitle = "Findings of the Association for Computational Linguistics: ACL 2025",
month = jul,
year = "2025",
address = "Vienna, Austria",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2025.findings-acl.119/",
doi = "10.18653/v1/2025.findings-acl.119",
pages = "2353--2366",
ISBN = "979-8-89176-256-5",
abstract = "To enhance the interpretability of multimodal unified representations, many studies have focused on discrete unified representations. These efforts typically start with contrastive learning and gradually extend to the disentanglement of modal information, achieving solid multimodal discrete unified representations. However, existing research often overlooks two critical issues: 1) The use of Euclidean distance for quantization in discrete representations often overlooks the important distinctions among different dimensions of features, resulting in redundant representations after quantization; 2) Different modalities have unique characteristics, and a uniform alignment approach does not fully exploit these traits. To address these issues, we propose Training-free Optimization of Codebook (TOC) and Fine and Coarse cross-modal Information Disentangling (FCID). These methods refine the unified discrete representations from pretraining and perform fine- and coarse-grained information disentanglement tailored to the specific characteristics of each modality, achieving significant performance improvements over previous state-of-the-art models. The code is available at https://github.com/haihuangcode/CMG."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="huang-etal-2025-enhancing-multimodal">
<titleInfo>
<title>Enhancing Multimodal Unified Representations for Cross Modal Generalization</title>
</titleInfo>
<name type="personal">
<namePart type="given">Hai</namePart>
<namePart type="family">Huang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yan</namePart>
<namePart type="family">Xia</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Shengpeng</namePart>
<namePart type="family">Ji</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Shulei</namePart>
<namePart type="family">Wang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Hanting</namePart>
<namePart type="family">Wang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Minghui</namePart>
<namePart type="family">Fang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jieming</namePart>
<namePart type="family">Zhu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Zhenhua</namePart>
<namePart type="family">Dong</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Sashuai</namePart>
<namePart type="family">Zhou</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Zhou</namePart>
<namePart type="family">Zhao</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2025-07</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Findings of the Association for Computational Linguistics: ACL 2025</title>
</titleInfo>
<name type="personal">
<namePart type="given">Wanxiang</namePart>
<namePart type="family">Che</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Joyce</namePart>
<namePart type="family">Nabende</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Ekaterina</namePart>
<namePart type="family">Shutova</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Mohammad</namePart>
<namePart type="given">Taher</namePart>
<namePart type="family">Pilehvar</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Vienna, Austria</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
<identifier type="isbn">979-8-89176-256-5</identifier>
</relatedItem>
<abstract>To enhance the interpretability of multimodal unified representations, many studies have focused on discrete unified representations. These efforts typically start with contrastive learning and gradually extend to the disentanglement of modal information, achieving solid multimodal discrete unified representations. However, existing research often overlooks two critical issues: 1) The use of Euclidean distance for quantization in discrete representations often overlooks the important distinctions among different dimensions of features, resulting in redundant representations after quantization; 2) Different modalities have unique characteristics, and a uniform alignment approach does not fully exploit these traits. To address these issues, we propose Training-free Optimization of Codebook (TOC) and Fine and Coarse cross-modal Information Disentangling (FCID). These methods refine the unified discrete representations from pretraining and perform fine- and coarse-grained information disentanglement tailored to the specific characteristics of each modality, achieving significant performance improvements over previous state-of-the-art models. The code is available at https://github.com/haihuangcode/CMG.</abstract>
<identifier type="citekey">huang-etal-2025-enhancing-multimodal</identifier>
<identifier type="doi">10.18653/v1/2025.findings-acl.119</identifier>
<location>
<url>https://aclanthology.org/2025.findings-acl.119/</url>
</location>
<part>
<date>2025-07</date>
<extent unit="page">
<start>2353</start>
<end>2366</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Enhancing Multimodal Unified Representations for Cross Modal Generalization
%A Huang, Hai
%A Xia, Yan
%A Ji, Shengpeng
%A Wang, Shulei
%A Wang, Hanting
%A Fang, Minghui
%A Zhu, Jieming
%A Dong, Zhenhua
%A Zhou, Sashuai
%A Zhao, Zhou
%Y Che, Wanxiang
%Y Nabende, Joyce
%Y Shutova, Ekaterina
%Y Pilehvar, Mohammad Taher
%S Findings of the Association for Computational Linguistics: ACL 2025
%D 2025
%8 July
%I Association for Computational Linguistics
%C Vienna, Austria
%@ 979-8-89176-256-5
%F huang-etal-2025-enhancing-multimodal
%X To enhance the interpretability of multimodal unified representations, many studies have focused on discrete unified representations. These efforts typically start with contrastive learning and gradually extend to the disentanglement of modal information, achieving solid multimodal discrete unified representations. However, existing research often overlooks two critical issues: 1) The use of Euclidean distance for quantization in discrete representations often overlooks the important distinctions among different dimensions of features, resulting in redundant representations after quantization; 2) Different modalities have unique characteristics, and a uniform alignment approach does not fully exploit these traits. To address these issues, we propose Training-free Optimization of Codebook (TOC) and Fine and Coarse cross-modal Information Disentangling (FCID). These methods refine the unified discrete representations from pretraining and perform fine- and coarse-grained information disentanglement tailored to the specific characteristics of each modality, achieving significant performance improvements over previous state-of-the-art models. The code is available at https://github.com/haihuangcode/CMG.
%R 10.18653/v1/2025.findings-acl.119
%U https://aclanthology.org/2025.findings-acl.119/
%U https://doi.org/10.18653/v1/2025.findings-acl.119
%P 2353-2366
Markdown (Informal)
[Enhancing Multimodal Unified Representations for Cross Modal Generalization](https://aclanthology.org/2025.findings-acl.119/) (Huang et al., Findings 2025)
ACL
- Hai Huang, Yan Xia, Shengpeng Ji, Shulei Wang, Hanting Wang, Minghui Fang, Jieming Zhu, Zhenhua Dong, Sashuai Zhou, and Zhou Zhao. 2025. Enhancing Multimodal Unified Representations for Cross Modal Generalization. In Findings of the Association for Computational Linguistics: ACL 2025, pages 2353–2366, Vienna, Austria. Association for Computational Linguistics.