Non-Parametric Approximate Linear Programming for MDPs.
Jason PazisRonald ParrPublished in: AAAI (2011)
Keyphrases
- linear programming
- markov decision problems
- factored mdps
- markov decision processes
- dynamic programming
- linear program
- factored markov decision processes
- reinforcement learning
- policy iteration
- average cost
- policy evaluation
- integer programming
- optimal solution
- nonlinear programming
- quadratic programming
- optimal policy
- quality guarantees
- constraint propagation
- network flow
- markov decision process
- decision theoretic planning
- objective function
- finite horizon
- primal dual
- state space
- np hard
- temporal difference
- transition probabilities
- planning under uncertainty
- exact solution
- finite state
- special case
- lower bound