The Geometry of Memoryless Stochastic Policy Optimization in Infinite-Horizon POMDPs.
Johannes MüllerGuido MontúfarPublished in: CoRR (2021)
Keyphrases
- infinite horizon
- optimal policy
- policy iteration algorithm
- partially observable
- markov decision processes
- partially observable markov decision processes
- dec pomdps
- finite horizon
- long run
- dynamic programming
- stochastic demand
- periodic review
- policy iteration
- markov decision process
- finite state
- optimal control
- state space
- control policies
- production planning
- state dependent
- average cost
- markov decision problems
- reinforcement learning
- decision problems
- single item
- stochastic programming
- continuous state
- holding cost
- fixed cost
- optimal production
- lead time
- average reward
- planning under uncertainty
- inventory level
- learning algorithm
- graphical models