Exploration in Approximate Hyper-State Space for Meta Reinforcement Learning.
Luisa M. ZintgrafLeo FengMaximilian IglKristian HartikainenKatja HofmannShimon WhitesonPublished in: CoRR (2020)
Keyphrases
- state space
- reinforcement learning
- reinforcement learning algorithms
- exploration strategy
- markov decision processes
- policy evaluation
- exploration exploitation
- optimal policy
- factored mdps
- function approximation
- action selection
- action space
- dynamic programming
- model based reinforcement learning
- heuristic search
- continuous state spaces
- markov decision process
- state variables
- markov chain
- state abstraction
- learning algorithm
- model free
- dynamical systems
- temporal difference
- active exploration
- particle filter
- partially observable
- machine learning
- macro actions
- control problems
- belief state
- state transition
- learning problems
- planning problems
- exploration exploitation tradeoff
- supervised learning
- finite state
- exact solution
- markov decision problems
- reinforcement learning methods
- reward function
- autonomous learning
- policy search
- meta level
- learning tasks
- learning process
- multi agent
- neural network
- learning capabilities