提出MuMo框架，提升多模态分子表示的稳定性和泛化能力
Structure-Aware Fusion with Progressive Injection for Multimodal Molecular Representation Learning
- 设计结构融合管道，整合2D拓扑与3D几何构建稳定结构先验
- 采用渐进式注入机制，在序列流中不对称融合先验，避免模态坍塌
- 在29个任务中平均提升2.7%，22个任务排名第一，尤其在LD50任务提升27%
多模态分子模型常受三维构象不可靠和模态坍塌影响，限制其鲁棒性与泛化能力。本文提出MuMo，一种结构感知的多模态融合框架，通过两项关键策略解决该问题：为降低依赖构象的融合不稳定性，设计结构融合管道（SFP），将2D拓扑与3D几何融合为统一且稳定的结构先验；为缓解朴素融合引发的模态坍塌，引入渐进式注入（PI）机制，将该先验非对称地融入序列流，保留模态特异性建模的同时实现跨模态增强。基于状态空间主干网络，MuMo支持长距离依赖建模与鲁棒信息传播。在来自Therapeutics Data Commons（TDC）和MoleculeNet的29个基准任务上，MuMo在每个任务上均超越最优基线平均2.7%，在22个任务中排名第一，其中在LD50任务提升27%。结果验证了其对3D构象噪声的鲁棒性及多模态融合的有效性。代码已开源：github.com/selmiss/MuMo。
原文摘要 · Abstract (English)
Multimodal molecular models often suffer from 3D conformer unreliability and modality collapse, limiting their robustness and generalization. We propose MuMo, a structured multimodal fusion framework that addresses these challenges in molecular representation through two key strategies. To reduce the instability of conformer-dependent fusion, we design a Structured Fusion Pipeline (SFP) that combines 2D topology and 3D geometry into a unified and stable structural prior. To mitigate modality collapse caused by naive fusion, we introduce a Progressive Injection (PI) mechanism that asymmetrically integrates this prior into the sequence stream, preserving modality-specific modeling while enabling cross-modal enrichment. Built on a state space backbone, MuMo supports long-range dependency modeling and robust information propagation. Across 29 benchmark tasks from Therapeutics Data Commons (TDC) and MoleculeNet, MuMo achieves an average improvement of 2.7% over the best-performing baseline on each task, ranking first on 22 of them, including a 27% improvement on the LD50 task. These results validate its robustness to 3D conformer noise and the effectiveness of multimodal fusion in molecular representation. The code is available at: github.com/selmiss/MuMo.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。