Reinforcement Learning in Nonzero-sum Linear Quadratic Deep Structured Games: Global Convergence of Policy Optimization.
Masoud RoudneshinJalal ArabneydiAmir G. AghdamPublished in: CDC (2020)
Keyphrases
- global convergence
- reinforcement learning
- optimization methods
- global optimum
- convergence analysis
- optimal policy
- convergence rate
- convergence speed
- linear quadratic
- optimal control
- optimization method
- globally convergent
- convex minimization
- line search
- closed loop
- optimization algorithm
- objective function
- dynamic programming
- real time
- learning problems
- hybrid algorithm
- particle swarm
- vector valued
- optimization problems
- optimal solution
- dynamical systems
- metaheuristic
- step size
- maximum likelihood
- special case
- learning algorithm