Skill-based Meta-Reinforcement Learning.
Taewook NamShao-Hua SunKarl PertschSung Ju HwangJoseph J. LimPublished in: ICLR (2022)
Keyphrases
- reinforcement learning
- function approximation
- state space
- optimal policy
- robotic control
- policy search
- learning algorithm
- direct policy search
- control problems
- machine learning
- stochastic approximation
- real time
- behavioural cloning
- data sets
- neural network
- markov decision processes
- transfer learning
- multi agent
- optimal control
- information systems
- temporal difference learning
- dynamic programming
- skill acquisition
- transition model
- relational reinforcement learning
- skills needed
- high level abstraction
- artificial neural networks