Fully Distributed Actor-Critic Architecture for Multitask Deep Reinforcement Learning.
Sergio Valcarcel MacuaIan DaviesAleksi TukiainenEnrique Munoz de CotePublished in: CoRR (2021)
Keyphrases
- actor critic
- fully distributed
- reinforcement learning
- multi task
- policy gradient
- temporal difference
- learning problems
- optimal control
- transfer learning
- reinforcement learning algorithms
- approximate dynamic programming
- loosely coupled
- function approximation
- cooperative
- learning tasks
- neuro fuzzy
- peer to peer
- gradient method
- overlay network
- multi agent systems
- learning algorithm
- machine learning
- feature selection
- multi agent
- policy iteration
- multi class
- model free
- data mining
- average reward
- rl algorithms
- action selection
- reinforcement learning methods
- distributed search
- temporal difference learning
- dynamic programming
- step size
- state space