Scalable solutions of interactive POMDPs using generalized and bounded policy iteration.
Ekhlas SonuPrashant DoshiPublished in: Auton. Agents Multi Agent Syst. (2015)
Keyphrases
- policy iteration
- markov decision processes
- reinforcement learning
- optimal policy
- markov decision problems
- policy iteration algorithm
- finite state
- model free
- average reward
- markov decision process
- sample path
- state space
- fixed point
- infinite horizon
- policy evaluation
- dynamic programming
- optimal control
- optimal solution
- partially observable
- partially observable markov decision processes
- reinforcement learning algorithms
- least squares
- belief state
- actor critic
- temporal difference
- neural network