Methods overview

This section introduces the recurring model families and learning paradigms used throughout the decision framework. Together, these pages provide a high-level overview of where different method families are most useful, how they are typically applied, and what kinds of data or practical constraints they are designed to address.

The method families covered in the current paper are:

• Foundation models, self-supervised learning and transfer learning
• Physics-informed and hybrid models
• Sequence and dynamics-based models
• Graph-based methods, geometric models and unstructured meshes
• Models for explainability and uncertainty
• Generative models and synthetic augmentation
• Reinforcement learning

These pages are adapted directly from the paper text.