Exploration in Approximate Hyper-State Space for Meta Reinforcement Learning.
Luisa M. ZintgrafLeo FengCong LuMaximilian IglKristian HartikainenKatja HofmannShimon WhitesonPublished in: ICML (2021)
Keyphrases
- state space
- reinforcement learning
- reinforcement learning algorithms
- exploration strategy
- markov decision processes
- optimal policy
- factored mdps
- policy evaluation
- exploration exploitation
- action selection
- function approximation
- model based reinforcement learning
- continuous state spaces
- dynamic programming
- state abstraction
- action space
- reinforcement learning methods
- heuristic search
- markov chain
- active exploration
- multi agent
- particle filter
- partially observable
- temporal difference
- control problems
- learning algorithm
- dynamical systems
- meta level
- state variables
- markov decision process
- supervised learning
- autonomous learning
- multi agent systems
- goal state
- continuous state
- planning problems
- reward shaping
- transfer learning
- neural network
- policy search
- initial state
- reward function
- sufficient conditions
- real valued
- domain independent