Derivative-Informed Projected Neural Networks for High-Dimensional Parametric Maps Governed by PDEs.
Thomas O'Leary-RoseberryUmberto VillaPeng ChenOmar GhattasPublished in: CoRR (2020)
Keyphrases
- high dimensional
- neural network
- partial differential equations
- low dimensional
- nearest neighbor
- level set
- dimensionality reduction
- parameter space
- high dimensional data
- back propagation
- similarity search
- multi dimensional
- pattern recognition
- sparse data
- fuzzy logic
- data points
- artificial neural networks
- self organizing maps
- image processing
- neural nets
- fault diagnosis
- structured light
- fuzzy systems
- multi layer
- high dimensionality
- data sets
- feature space
- computer vision
- multilayer perceptron
- neural network model
- variable selection
- training process
- feature maps
- numerical solution
- discrete data
- high dimensional problems
- projector camera