Performance Issues for the Iterative Solution of Markov Decision Processes on Parallel Computers.
Thomas W. ArchibaldK. I. M. McKinnonLyn C. ThomasPublished in: INFORMS J. Comput. (1995)
Keyphrases
- markov decision processes
- finite state
- parallel computers
- state space
- optimal policy
- policy iteration
- reinforcement learning
- dynamic programming
- transition matrices
- decision theoretic planning
- parallel computing
- planning under uncertainty
- action space
- markov decision process
- model based reinforcement learning
- infinite horizon
- average reward
- average cost
- optimal solution
- partially observable
- computer architecture
- parallel implementation
- fixed point
- data transfer
- reward function
- massively parallel
- numerical methods
- action sets
- parallel processing