Explore then Execute: Adapting without Rewards via Factorized Meta-Reinforcement Learning.
Evan Zheran LiuAditi RaghunathanPercy LiangChelsea FinnPublished in: CoRR (2020)
Keyphrases
- reinforcement learning
- markov decision processes
- function approximation
- model free
- state space
- learning algorithm
- reinforcement learning algorithms
- optimal policy
- machine learning
- reward function
- meta level
- learning problems
- transfer learning
- control policy
- reward shaping
- real time
- data sets
- optimal control
- matrix factorization
- action selection
- temporal difference
- supervised learning
- dynamic programming
- learning capabilities
- partially observable
- learning process
- reinforcement learning methods
- hidden state
- meta reasoning
- multi agent
- robotic control