An SMDP approach for Reinforcement Learning in HPC cluster schedulers.
Renato Luiz de Freitas CunhaLuiz ChaimowiczPublished in: Future Gener. Comput. Syst. (2023)
Keyphrases
- reward shaping
- reinforcement learning
- semi markov decision process
- markov decision problems
- markov decision processes
- semi markov decision processes
- reinforcement learning algorithms
- state and action spaces
- complex domains
- state space
- average reward
- optimal policy
- model free
- function approximation
- markov decision process
- hierarchical reinforcement learning
- dynamic programming
- learning algorithm
- action selection
- temporal difference
- grid computing
- action space
- optimal control
- state abstraction
- data points