Neural-network-based discounted optimal control via an integrated value iteration with accuracy guarantee.
Mingming HaDing WangDerong LiuPublished in: Neural Networks (2021)
Keyphrases
- optimal control
- infinite horizon
- dynamic programming
- markov decision processes
- finite horizon
- control problems
- risk sensitive
- feedback control
- average cost
- policy iteration
- partially observable
- neural network
- state space
- control strategy
- reinforcement learning
- markov decision process
- optimal control problems
- class of nonlinear systems
- optimal policy
- brownian motion
- production planning
- long run
- knapsack problem
- average reward
- lyapunov function
- linear programming