Deep neural networks based temporal-difference methods for high-dimensional parabolic partial differential equations.
Shaojie ZengYihua CaiQingsong ZouPublished in: J. Comput. Phys. (2022)
Keyphrases
- partial differential equations
- high dimensional
- neural network
- temporal difference methods
- function approximators
- function approximation
- level set
- image processing
- image denoising
- numerical solution
- temporal difference
- image enhancement
- pattern recognition
- multiscale
- reinforcement learning problems
- policy search
- high order
- artificial neural networks
- evolutionary methods
- feature space
- data points
- policy evaluation
- reinforcement learning
- reinforcement learning algorithms
- genetic algorithm
- learning tasks
- markov decision processes