A Sampling-Based Approach to Reducing the Complexity of Continuous State Space POMDPs by Decomposition Into Coupled Perceptual and Decision Processes.
Rasool FakoorManfred HuberPublished in: ICMLA (1) (2012)
Keyphrases
- decision processes
- continuous state spaces
- decision problems
- markov decision processes
- partially observable markov decision processes
- markov decision problems
- state space
- continuous state
- reinforcement learning
- partially observable
- decision process
- computational complexity
- action space
- finite state
- decision making
- optimal policy
- reasoning process
- belief state
- markov chain
- dynamic programming
- policy iteration
- training data
- real valued
- machine learning
- dynamical systems
- np hard
- multi agent