Symbolic Dynamic Programming for First-Order MDPs.
Craig BoutilierRaymond ReiterBob PricePublished in: IJCAI (2001)
Keyphrases
- dynamic programming
- markov decision processes
- decision diagrams
- state space
- optimal policy
- binary decision diagrams
- dec pomdps
- markov decision problems
- reinforcement learning
- higher order
- dynamic programming algorithms
- finite state
- first order logic
- infinite horizon
- finite horizon
- stereo matching
- decision theoretic planning
- linear programming
- policy iteration
- optimal control
- greedy algorithm
- factored mdps
- multi valued
- planning under uncertainty
- policy search
- high level
- real time dynamic programming
- average cost
- efficient computation
- symbolic representation
- multi agent
- markov decision process
- partially observable
- decision theoretic
- boolean functions
- sufficient conditions
- markov chain
- state and action spaces