A Smart Cache Content Update Policy Based on Deep Reinforcement Learning.
Lincan LiChiew Foong KwongQianyu LiuJing WangPublished in: Wirel. Commun. Mob. Comput. (2020)
Keyphrases
- reinforcement learning
- optimal policy
- cache management
- policy search
- markov decision process
- action selection
- replacement policy
- dynamic content
- policy gradient
- control policies
- policy evaluation
- approximate dynamic programming
- reinforcement learning algorithms
- reward function
- model free
- markov decision problems
- reinforcement learning problems
- partially observable environments
- partially observable
- data access
- multimedia
- state and action spaces
- dynamic programming
- actor critic
- action space
- state space
- markov decision processes
- continuous state spaces
- distributed object
- partially observable domains
- policy iteration
- hit rate
- state action
- average reward
- content delivery
- machine learning
- temporal difference
- function approximation
- multimedia content
- main memory
- query processing
- metadata
- learning algorithm