Optimality conditions for total-cost Partially Observable Markov Decision Processes.
Eugene A. FeinbergPavlo O. KasyanovMichael Z. ZgurovskyPublished in: CDC (2013)
Keyphrases
- total cost
- optimality conditions
- partially observable markov decision processes
- finite state
- reinforcement learning
- nonlinear programming
- decision problems
- dynamical systems
- optimal policy
- dynamic programming
- belief state
- markov decision processes
- lower level
- state space
- planning problems
- optimal solution
- multi agent
- lead time
- linear programming
- infinite horizon
- sample size
- markov chain
- decision making
- mathematical programming
- fixed point