Similarities between policy gradient methods in reinforcement and supervised learning.
Eric BenhamouDavid SaltielPublished in: ESANN (2020)
Keyphrases
- supervised learning
- reinforcement learning
- policy gradient methods
- natural actor critic
- policy gradient
- actor critic
- function approximation
- reinforcement learning methods
- learning algorithm
- active learning
- training data
- reinforcement learning algorithms
- temporal difference
- machine learning
- state space
- learning tasks
- learning problems
- model free
- robot arm
- dynamic programming
- reinforcement learning problems
- semi supervised
- optimal control
- control problems
- labeled data
- approximate dynamic programming
- control strategies
- linear programming
- decision trees