An adaptive power management with policy selection mechanism based on reward-punishment scheme.
Fa-Gui LiuXiao-Yong XingJin-Biao LinJun LinPublished in: ICMLC (2013)
Keyphrases
- selection mechanism
- power management
- power consumption
- agent receives
- reinforcement learning
- selective attention
- reward function
- energy consumption
- energy saving
- data center
- average reward
- energy efficiency
- optimal policy
- markov decision process
- state space
- learning environment
- policy gradient
- real time
- long run
- low cost
- action selection
- sensor networks
- dynamic programming
- multi agent
- partially observable environments
- databases