Value function optimistic initialization with uncertainty and confidence awareness in lifelong reinforcement learning.
Soumia MehimehXianglong TangWei ZhaoPublished in: Knowl. Based Syst. (2023)
Keyphrases
- reinforcement learning
- learning algorithm
- state space
- neural network
- uncertain data
- machine learning
- partial observability
- function approximators
- supervised learning
- markov decision processes
- dynamic programming
- function approximation
- belief functions
- reinforcement learning algorithms
- confidence level
- learning process
- mobile devices
- policy gradient
- confidence values
- transition model