Into the danger zone: stable extrapolation in high-dimensional function and operator learning
math.NA, cs.LG, cs.NA, stat.ML
Submitted: 2026-09-29
Updated: 2026-10-03
Terminology
Sources
- The Sample Complexity of Learning Lipschitz Operators with respect to Gaussian Measures
- Out of Distribution Generalization in Machine Learning
- A short tour of operator learning theory: Convergence rates, statistical limits, and open questions
- Defining Foundation Models for Computational Science: A Call for Clarity and Rigor
- Do Physics Foundation Models Learn Generalizable Physics? A Bias-Aware Benchmark Across Physical Regimes and Distribution Shifts
- Optimal Recovery Meets Minimax Estimation
- Understanding and Mitigating Extrapolation Failures in Physics-Informed Neural Networks
- Learning where to learn: Training data distribution optimization for scientific machine learning
- Bridging the Gap Between Approximation and Learning via Optimal Approximation by ReLU MLPs of Maximal Regularity
- Algorithmic Foundations of Deep Learning: Complexity-Theoretic Rates and a Characterization of Universal Approximation
- Error estimates for DeepOnets: A deep learning framework in infinite dimensions
- The Parametric Complexity of Operator Learning
- Towards Out-Of-Distribution Generalization: A Survey
- Out-of-distribution generalization of deep-learning surrogates for 2D PDE-generated dynamics in the small-data regime
- Statistical Learning Theory for Neural Operators
- Test-time Generalization for Physics through Neural Operator Splitting
- SIMSHIFT: A Benchmark for Adapting Neural Surrogates to Distribution Shifts
- Diagnosing Failure Modes of Neural Operators Across Diverse PDE Families
- Is Zero-Shot Super-Resolution Possible in Operator Learning?
- Performance of Neural and Polynomial Operator Surrogates
Related papers
- Do physics-informed neural networks (PINNs) need to be deep? Shallow PINNs using the Levenberg-Marquardt algorithm
- A Neural-preconditioned Poisson Solver for Mixed Dirichlet and Neumann Boundary Conditions
- Second-order consistency for learning chaotic dynamics via randomized Jacobian matching
- Windowed thinning and query complexity for the bouncy particle and Zigzag samplers
- Data-efficient Kernel Methods for Learning Hamiltonian Systems
- Adjoint Method versus Physics-Informed Neural Networks in PDE-Constrained Inverse Problems