A Meta-MDP Approach to Exploration for Lifelong Reinforcement Learning.
Francisco M. GarciaPhilip S. ThomasPublished in: AAMAS (2019)
Keyphrases
- reinforcement learning
- exploration strategy
- markov decision processes
- model based reinforcement learning
- markov decision process
- optimal policy
- exploration exploitation
- action selection
- state space
- active exploration
- function approximation
- reinforcement learning algorithms
- markov decision problems
- partially observable
- unknown environments
- reward function
- action space
- state and action spaces
- action sets
- policy iteration
- temporal difference
- dynamic programming
- balancing exploration and exploitation
- exploration exploitation tradeoff
- finite state
- meta level
- multi agent
- reinforcement learning methods
- competence development
- lifelong learning
- model free
- st century
- learning activities
- learning algorithm
- average reward
- partially observable markov decision processes
- machine learning
- continuous state
- factored mdps
- evaluation function
- decision problems
- linear program
- markov chain
- supervised learning
- bayesian reinforcement learning
- learning process