MQGrad: Reinforcement Learning of Gradient Quantization in Parameter Server.
Guoxin CuiJun XuWei ZengYanyan LanJiafeng GuoXueqi ChengPublished in: CoRR (2018)
Keyphrases
- reinforcement learning
- function approximation
- client server
- policy gradient
- web server
- database
- state space
- input parameters
- edge detection
- machine learning
- reinforcement learning algorithms
- parameter settings
- multi agent
- parameter values
- temporal difference
- single server
- optimal policy
- model free
- low latency
- computational complexity
- action selection
- quantization error
- network bandwidth
- gradient information
- multiscale
- robotic control