Discrete and Continuous ambush games: optimal policies and approximate solutions.
Emmanuel BoidotAude MarzuoliEric FeronPublished in: CoRR (2016)
Keyphrases
- approximate solutions
- optimal policy
- markov decision processes
- partially observable markov decision processes
- state space
- finite horizon
- np hard
- optimal solution
- continuous variables
- multistage
- decision problems
- reinforcement learning
- exact solution
- long run
- sufficient conditions
- dynamic programming
- state dependent
- finite state
- average reward reinforcement learning
- average reward
- infinite horizon
- action space
- markov decision process
- semi markov decision processes
- serial inventory systems
- policy iteration
- energy function
- lost sales
- control policies
- initial state
- dynamic programming algorithms
- reward function
- image segmentation
- finite number
- monte carlo tree search
- higher order