Global Convergence of Policy Gradient Methods for the Linear Quadratic Regulator.
Maryam FazelRong GeSham M. KakadeMehran MesbahiPublished in: ICML (2018)
Keyphrases
- linear quadratic
- global convergence
- policy gradient methods
- optimal control
- policy gradient
- natural actor critic
- closed loop
- global optimum
- convergence rate
- vector valued
- dynamical systems
- optimization methods
- convergence speed
- gaussian model
- control system
- reinforcement learning
- particle swarm
- dynamic programming
- gradient method
- function approximation
- control strategy
- pso algorithm
- particle swarm optimization
- step size
- real valued
- computational complexity
- search algorithm
- optimal solution