Active Policy Iteration: Efficient Exploration through Active Learning for Value Function Approximation in Reinforcement Learning.
Takayuki AkiyamaHirotaka HachiyaMasashi SugiyamaPublished in: IJCAI (2009)
Keyphrases
- policy iteration
- reinforcement learning
- active learning
- temporal difference
- markov decision processes
- model free
- approximate dynamic programming
- temporal difference learning
- optimal policy
- policy evaluation
- function approximation
- markov decision process
- state space
- markov games
- supervised learning
- learning algorithm
- stochastic approximation
- reinforcement learning algorithms
- fixed point
- sample path
- least squares
- learning process
- machine learning
- average reward
- actor critic
- markov decision problems
- transfer learning
- semi supervised
- finite state
- evaluation function
- optimal control
- action selection
- state action
- monte carlo
- training set
- approximate policy iteration
- average cost
- step size
- action space
- control policy
- infinite horizon
- labeled data
- dynamic programming
- cost function
- state and action spaces