IMD-SMART DIAGNOSTICS

Interpolating multimodal data for smart diagnostics and inference - Applied intelligence for decision-making in a dynamic system

This project faces several major scientific challenges. Firstly, the heterogeneity of multimodal [11] data poses a major challenge. The aim is to achieve coherent integration of data from different modalities, each with different structures and formats. This challenge calls for innovative solutions to ensure the compatibility and convergence of information from different sources. Learning with limited data [10] is another crucial challenge. The difficulty lies in obtaining the costly annotated data essential for training multimodal models, underlining the need to develop efficient approaches in a context of limited resources. Last but not least, model robustness and reliability are also key issues. Model reliability is often compromised in the face of variations and disturbances in industrial environments, requiring innovative strategies to guarantee stable and consistent performance despite changing conditions.