Meta Reinforcement Learning with Task Embedding and Shared Policy.
Lin LanZhenguo LiXiaohong GuanPinghui WangPublished in: CoRR (2019)
Keyphrases
- reinforcement learning
- optimal policy
- policy search
- markov decision process
- action selection
- partially observable
- state space
- control policy
- function approximators
- policy gradient
- reinforcement learning problems
- markov decision problems
- continuous state
- reinforcement learning algorithms
- markov decision processes
- control policies
- actor critic
- reward function
- partially observable environments
- state and action spaces
- approximate dynamic programming
- action space
- meta level
- policy evaluation
- function approximation
- dynamic programming
- agent learns
- rl algorithms
- temporal difference
- decision problems
- policy iteration
- vector space
- learning algorithm
- average reward
- state action
- partially observable markov decision processes
- graph embedding
- state dependent
- model free
- supervised learning
- neural network
- infinite horizon
- long run
- partially observable domains
- optimal control
- continuous state spaces
- transition model
- model free reinforcement learning
- exploration exploitation tradeoff
- agent receives
- inverse reinforcement learning
- policy making
- nonlinear dimensionality reduction
- finite state
- transfer learning
- multi agent
- machine learning