ACL Findings2026
SCOPE: Preserving Modality-Specific Cues to Mitigate Modality Laziness in Multimodal Learning
Abstract
Multimodal learning aims to learn unified multimodal representations from heterogeneous modalities and supports many natural language processing tasks. However, multimodal models often exhibit modality laziness: over-relying on a dominant modality and under-exploiting complementary signals. Existing approaches typically strengthen unimodal training or rebalance modality contributions, but they may still emphasize shared semantics and overlook modality-specific cues. To address this, we propose SCOPE, a unified framework for learning complete multimodal representations, achieving Shared-and-COmplementary cue PrEservation. Firstly, SCOPE uses a mutual informationguided disentanglement module to separate shared semantics from modality-specific cues and mitigate representation collapse. Secondly, SCOPE aligns modalities by enforcing structural consistency between modality-wise semantic graphs, avoiding brittle point-wise matching. Finally, SCOPE performs balanced fusion via structure-aware diffusion attention to integrate shared and complementary cues without feature homogenization. Experiments on five benchmark datasets show that SCOPE consistently outperforms SOTA baselines, achieving up to 27.10% accuracy improvement.
Ask Lune
Know this paper in minutes, not an afternoon.
Lune can read every page of it and the top-tier papers that cite it, and gives you its claims, its evidence and its limits, each one cited to the paper it comes from. No skimming, no AI slop.
Lineage
Builds on26
- MISA: Modality-Invariant and -Specific Representations for Multimodal Sentiment AnalysisDevamanyu Hazarika, Roger Zimmermann, Soujanya PoriaACM MM 2020 · 1,037 citations
- Balanced Multimodal Learning via On-the-fly Gradient ModulationXiaokang Peng, Yake Wei, Andong Deng, Dong Wang et al.CVPR 2022 · 264 citations
- Modality Competition: What Makes Joint Training of Multi-modal Network Fail in Deep Learning? (Provably)Yu Huang, Junyang Lin, Chang Zhou, Hongxia Yang et al.ICML 2022 · 168 citations
- Are Multimodal Transformers Robust to Missing Modality?Mengmeng Ma, Jian Ren, Long Zhao, Davide Testuggine et al.CVPR 2022 · 153 citations
- ConFEDE: Contrastive Feature Decomposition for Multimodal Sentiment AnalysisJiuding Yang, Yakun Yu, Di Niu, Weidong Guo et al.ACL 2023 · 135 citations
Related papers
- Multi-View Differential Mixing and Graph-Guided Structural Region Selection for Cross-Modal AlignmentLinlin Ji, Li LiuAAAI 2026
- DecAlign: Hierarchical Cross-Modal Alignment for Decoupled Multimodal Representation LearningChengxuan Qian, Shuo Xing, Li Li, Yue Zhao et al.ICLR 2026 · 42 citations
- MMOne: Representing Multiple Modalities in One SceneZhifeng Gu, Bing WangICCV 2025
- Structures Meet Semantics: Multimodal Fusion via Graph Contrastive LearningJiangfeng Sun, Sihao He, Zhonghong Ou, Meina SongAAAI 2026
- DLF: Disentangled-Language-Focused Multimodal Sentiment AnalysisPan Wang, Qiang Zhou, Yawen Wu, Tianlong Chen et al.AAAI 2025 · 84 citations
