International workshop on Deep learning for Multi-modal Data
24-24 nov. 2025
Okinawa Institute of Science and Technology (OIST), Japan - Okinawa (Japon)
Multimodal data is everywhere, yet making sense of its complexity requires advancing deep learning far beyond traditional paradigms. This workshop brings together academic researchers and industry professionals to address core challenges in this field. We focus on deep learning techniques, particularly unsupervised, self-supervised and weakly supervised approaches, that learn transferable latent representations across modalities, moving beyond unimodal and static paradigms. Key topics include multi-view architectures, cross-modal alignment, attention mechanisms for modality fusion, diversity-aware ensemble learning, and explainable collaborative frameworks. Emphasis lies on adaptability to dynamic, incomplete, or context-dependent data structures, as well as scalable deployment. We encourage contributions on critical real-world applications such as health monitoring, autonomous systems, and environmental modeling. Featuring technical presentations, panel discussions and collaborative sessions, the workshop highlights theoretical advances and practical solutions. It fosters interdisciplinary dialogue on emerging challenges: computational efficiency, missing modality, evaluation standards and ethical considerations, charting future directions for deep multimodal learning.
Discipline scientifique :
Intelligence artificielle
Lieu de la conférence