Efficient Solving of Markov Decision Processes on GPUs Using Parallelized Sparse Matrices.
Adrian E. SapioShuvra S. BhattacharyyaMarilyn WolfPublished in: DASIP (2018)
Keyphrases
- fixed point
- sparse matrices
- floating point
- markov decision processes
- policy iteration
- transition matrices
- optimal policy
- markov decision problems
- semi markov decision processes
- finite state
- reinforcement learning
- state space
- decision theoretic planning
- stochastic shortest path
- dynamic programming
- infinite horizon
- factored mdps
- markov decision process
- average reward
- average cost
- condition number
- linear combination
- optical flow
- lower bound