analiza aiqbench
This arXiv survey reviews how Mixture-of-Experts architectures are applied to multimodal learning, organizing the literature around three roles: efficient scaling, representation learning, and adapting to imperfect data such as modality imbalance and missing modalities. It is a literature review rather than a new result, so its value lies in the taxonomy and the open problems it flags, including interpretable routing, expert communication, and lifelong multimodal learning. Readers building large multimodal models should treat it as a map of the design space, not as evidence that any specific approach works better.
Tackling Multimodal Learning Challenges with Mixture-of-Expert: A Survey
arXiv:2605.27431v2 Announce Type: replace-cross Abstract: Mixture-of-Experts (MoE) presents a naturally compatible and scalable framework for multimodal learning, demonstrating strong adaptability across diverse modalities and tasks. Despite its growing success, a comprehensive and systematic review on the MoE metho addressing multimodal challenges remains lacking. Existing surveys tend to evaluate either multimodal learning or MoE independently from method taxonomy, overlooking the unique interplay between them. This survey fills that gap by answering a central question: \textit{How does MoE effectively resolve multimodal challenges?} We approach this from three key perspectives: (1) \textbf{MoE as an Efficient Multimodal Engine:} enabling scalable multimodal modeling by decoupling computational cost from parameter growth and mitigating modality redundancy through selective expert activation; (2) \textbf{MoE as a Multimodal Representation Learner:} integrating complementary multi-opinion expert knowledge to enrich alignment and interaction representations; and (3) \textbf{MoE as a Multimodal Adapter:} providing a modular and flexible mechanism to model imperfect data scenarios such as modality imbalance and missing modality. Through our extensive literature review, we identify critical research gaps, including interpretable routing, expert communication, modality integration, and lifelong multimodal learning. We position this survey as a foundation for future research toward interpretable and sustainable multimodal Mixture-of-Experts system.
Abstract:Mixture-of-Experts (MoE) presents a naturally compatible and scalable framework for multimodal learning, demonstrating strong adaptability across diverse modalities and tasks. Despite its growing success, a comprehensive and systematic review on the MoE metho addressing multimodal challenges remains lacking. Existing surveys tend to evaluate either multimodal learning or MoE independently from method taxonomy, overlooking the unique interplay between them. This survey fills that gap by answering a central question: \textit{How does MoE effectively resolve multimodal challenges?} We approach this from three key perspectives: (1) \textbf{MoE as an Efficient Multimodal Engine:} enabling scalable multimodal modeling by decoupling computational cost from parameter growth and mitigating modality redundancy through selective expert activation; (2) \textbf{MoE as a Multimodal Representation Learner:} integrating complementary multi-opinion expert knowledge to enrich alignment and interaction representations; and (3) \textbf{MoE as a Multimodal Adapter:} providing a modular and flexible mechanism to model imperfect data scenarios such as modality imbalance and missing modality. Through our extensive literature review, we identify critical research gaps, including interpretable routing, expert communication, modality integration, and lifelong multimodal learning. We position this survey as a foundation for future research toward interpretable and sustainable multimodal Mixture-of-Experts system.
This survey has been accepted by IJCAI 2026. Results were released by 30 April 2026. This latest replacement is an extension version of survey to the previous version, which has been submitted to ACM Computing Survey for peer-review
Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
arXiv:2605.27431 [cs.LG]
(or arXiv:2605.27431v2 [cs.LG] for this version)
Text integral sub licența deschisă a publicației · 241 cuvinte