Reasoning about MDPs as Transformers of Probability Distributions.
Vijay Anand KorthikantiMahesh ViswanathanGul AghaYoungMin KwonPublished in: QEST (2010)
Keyphrases
- probability distribution
- markov decision processes
- random variables
- initial state
- state space
- factored mdps
- reinforcement learning
- spatial reasoning
- qualitative reasoning
- dynamic programming
- optimal policy
- utility function
- conditional probabilities
- bayesian networks
- decision theoretic planning
- formal theory
- average reward
- partial discharge
- stochastic processes
- normal distribution
- markov decision process
- partially observable
- markov decision problems
- machine learning
- semi markov decision processes
- model based reinforcement learning
- context specific
- planning under uncertainty
- reward function
- neural network