Towards Exploiting Duality in Approximate Linear Programming for MDPs.
Dmitri A. DolgovEdmund H. DurfeePublished in: AAAI (2005)
Keyphrases
- linear programming
- markov decision problems
- factored mdps
- linear program
- markov decision processes
- dynamic programming
- average cost
- factored markov decision processes
- policy iteration
- primal dual
- policy evaluation
- optimal solution
- column generation
- reinforcement learning
- np hard
- objective function
- network flow
- integer programming
- nonlinear programming
- exact solution
- quality guarantees
- decision theoretic planning
- state space
- special case
- partially observable
- algorithm for linear programming
- quadratic programming
- finite state