AI-ordlista
Den kompletta ordlistan över AI
Deep Neural Operator
Neural network architecture learning mappings between function spaces, capable of generalizing to different discretizations and boundary conditions in physical simulation.
Reduced Order Model
Dimensionality reduction technique preserving the essential characteristics of the original system while significantly reducing the computational cost of simulations.
Polynomial Chaos Expansion
Spectral representation method approximating a system's response by a linear combination of orthogonal polynomials, effective for uncertainty propagation.
Active Learning for Surrogates
Adaptive sampling strategy intelligently selecting the most informative training points to improve the efficiency of surrogate models.
Multi-fidelity Modeling
Approach combining simulations of different accuracies and computational costs to build accurate and resource-efficient surrogate models.
Proper Orthogonal Decomposition
Modal decomposition method identifying the dominant modes of a physical system to build reduced models preserving the essential energy of the system.
Radial Basis Function
Interpolation function using radially symmetric functions to construct response surfaces approximating multidimensional data in simulation.
Metamodeling
Construction of simplified models (meta-models) capturing input-output relationships of complex simulations while drastically reducing computation times.
Response Surface Methodology
Collection de techniques statistiques et mathématiques développant des modèles approximatifs de surfaces de réponse pour l'optimisation et l'analyse de sensibilité.
Transfer Learning for Physics
Adaptation de modèles pré-entraînés sur des domaines physiques similaires pour accélérer l'apprentissage de nouveaux modèles de surrogation avec moins de données.