Adaptable automation with modular deep reinforcement learning and policy transfer.
Zohreh RazieiMohsen MoghaddamPublished in: Eng. Appl. Artif. Intell. (2021)
Keyphrases
- reinforcement learning
- optimal policy
- policy search
- transfer learning
- markov decision process
- action selection
- markov decision processes
- reinforcement learning problems
- state space
- policy gradient
- action space
- policy iteration
- state and action spaces
- approximate dynamic programming
- function approximators
- partially observable environments
- policy evaluation
- function approximation
- control policy
- decision problems
- state dependent
- control policies
- markov decision problems
- partially observable markov decision processes
- actor critic
- reinforcement learning algorithms
- reward function
- partially observable
- model free
- average reward
- temporal difference
- state action
- cross domain
- knowledge transfer
- robotic control
- deep learning
- dynamic programming
- continuous state
- infinite horizon
- transition model
- modular structure
- control problems
- learning algorithm
- asymptotically optimal
- learning tasks
- monte carlo
- inverse reinforcement learning
- multi agent
- rl algorithms
- finite state
- agent receives
- model free reinforcement learning