Dyna-like reinforcement learning based on accumulative and average rewards.
Kao-Shing HwangChia-Yue LoPublished in: SMC (2010)
Keyphrases
- reinforcement learning
- function approximation
- temporal difference
- temporal difference learning
- reinforcement learning methods
- rl algorithms
- markov decision processes
- function approximators
- model free
- state space
- control problems
- partially observable
- multi agent
- machine learning
- average cost
- reinforcement learning algorithms
- learning algorithm
- optimal policy
- learning process
- action selection
- finite state
- reward function
- optimal control
- learning problems
- transfer learning
- action space
- dynamic programming
- reward shaping
- standard deviation
- policy iteration
- supervised learning
- state and action spaces