Episodic Reinforcement Learning by Logistic Reward-Weighted Regression.
Daan WierstraTom SchaulJan PetersJürgen SchmidhuberPublished in: ICANN (1) (2008)
Keyphrases
- reinforcement learning
- function approximation
- regression model
- state space
- temporal difference
- eligibility traces
- learning algorithm
- linear regression
- reinforcement learning algorithms
- model free
- optimal policy
- logistic regression
- markov decision processes
- gaussian processes
- regression algorithm
- transfer learning
- machine learning
- reward function
- multi armed bandit
- logistic model
- survival data
- model selection
- multi agent
- action selection
- long run
- gaussian process
- policy iteration
- state action
- locally weighted
- dynamic programming
- learning process
- total reward
- neural network