Point-wise maximum approach to approximate dynamic programming.
Paul N. BeuchatJoseph WarringtonJohn LygerosPublished in: CDC (2017)
Keyphrases
- approximate dynamic programming
- pointwise
- linear program
- weighted sum
- reinforcement learning
- dynamic programming
- stochastic dynamic programming
- step size
- policy iteration
- control policy
- confidence intervals
- linear programming
- optimal solution
- average cost
- sufficient conditions
- special case
- computational complexity
- machine learning