Parametric PDE Control with Deep Reinforcement Learning and Differentiable L0-Sparse Polynomial Policies.
Nicolò BotteghiUrban FaselPublished in: CoRR (2024)
Keyphrases
- reinforcement learning
- control policy
- optimal policy
- control policies
- control problems
- robot control
- policy search
- markov decision process
- adaptive control
- optimal control
- function approximation
- partial differential equations
- control strategies
- control system
- state space
- neural network
- control method
- high dimensional
- objective function
- sparse data
- anisotropic diffusion
- hierarchical reinforcement learning
- scale spaces
- reinforcement learning agents
- fitted q iteration
- temporal difference
- finite state
- control strategy
- high order
- sufficient conditions
- level set
- multiscale
- learning algorithm