On the Convergence Rate for Stochastic Approximation in the Nonsmooth Setting.
Eunji LimPublished in: Math. Oper. Res. (2011)
Keyphrases
- convergence rate
- stochastic approximation
- policy iteration
- step size
- monte carlo
- convergence speed
- learning rate
- gradient method
- primal dual
- global convergence
- temporal difference learning
- reinforcement learning
- special case
- numerical stability
- faster convergence rate
- average reward
- theoretical guarantees
- neural network