On the Global Convergence Rates of Softmax Policy Gradient Methods.
Jincheng MeiChenjun XiaoCsaba SzepesváriDale SchuurmansPublished in: ICML (2020)
Keyphrases
- global convergence
- policy gradient methods
- global optimum
- optimization methods
- convergence analysis
- convergence speed
- natural actor critic
- convergence rate
- convex minimization
- policy gradient
- optimization method
- robot arm
- hybrid algorithm
- computational intelligence
- step size
- search space
- optimal solution
- learning algorithm
- machine learning
- particle swarm
- objective function