A Continuous Internal-State Controller for Partially Observable Markov Decision Processes.
Yuki TaniguchiTakeshi MoriShin IshiiPublished in: ICANN (1) (2008)
Keyphrases
- partially observable markov decision processes
- internal state
- continuous action
- finite state
- dynamical systems
- reinforcement learning
- belief state
- belief space
- planning under uncertainty
- optimal policy
- decision problems
- stochastic domains
- continuous state
- dynamic programming
- markov decision processes
- control system
- initial state
- partially observable stochastic games
- state space
- selective perception
- multi agent
- mental states
- markov chain
- partially observable domains
- partially observable markov
- approximate solutions
- policy gradient
- partially observable
- learning algorithm
- predictive state representations
- point based value iteration
- infinite horizon
- optimal control
- control strategy
- belief revision
- intelligent agents