Estimates on the generalization error of Physics Informed Neural Networks (PINNs) for approximating PDEs.
Siddhartha MishraRoberto MolinaroPublished in: CoRR (2020)
Keyphrases
- generalization error
- neural network
- unbiased estimator
- cross validation
- model selection
- active learning
- learning algorithm
- training error
- upper bound
- supervised learning
- sample complexity
- binary classification
- partial differential equations
- linear classifiers
- sample size
- classification error
- training data
- training set
- target function
- low error
- subspace information criterion
- learning machines
- back propagation
- generalization error bounds
- perceptron learning
- artificial neural networks
- genetic algorithm
- image processing
- neural network model
- neural nets
- data sets
- training process
- lower bound
- machine learning