Skill-Critic: Refining Learned Skills for Reinforcement Learning.
Ce HaoCatherine WeaverChen TangKenta KawamotoMasayoshi TomizukaWei ZhanPublished in: CoRR (2023)
Keyphrases
- reinforcement learning
- function approximation
- temporal difference
- actor critic
- reinforcement learning algorithms
- skills needed
- policy gradient
- communication skills
- skill development
- state space
- skill acquisition
- previously learned
- technical skills
- model free
- markov decision processes
- multi agent
- optimal control
- lifelong learning
- learning process
- reinforcement learning methods
- gradient method
- learning algorithm
- optimal policy
- supervised learning
- neural network
- learning environment
- computer skills
- job skills
- declarative knowledge
- learned knowledge
- function approximators
- action space
- policy iteration
- dynamic programming