Continuous Control in Deep Reinforcement Learning with Direct Policy Derivation from Q Network.
Aydar AkhmetzyanovRauf YagfarovSalimzhan GafurovMikhail OstaninAlexandr KlimchikPublished in: IHIET (Lausanne) (2020)
Keyphrases
- reinforcement learning
- control policy
- control problems
- control policies
- action selection
- optimal policy
- continuous state spaces
- action space
- policy search
- state space
- network traffic
- markov decision process
- robot control
- function approximation
- optimal control
- network structure
- machine learning
- control strategies
- network model
- control system
- computer networks
- markov decision processes
- state action
- robotic control
- stochastic control
- wireless sensor networks
- actor critic
- peer to peer
- neural network
- communication networks
- control strategy
- partially observable
- temporal difference
- policy evaluation
- approximate dynamic programming