Reinforcement Learning When All Actions are Not Always Available.
Yash ChandakGeorgios TheocharousBlossom MetevierPhilip S. ThomasPublished in: CoRR (2019)
Keyphrases
- reinforcement learning
- action selection
- perceptual aliasing
- function approximation
- partially observable
- partially observable domains
- action sets
- action space
- reward function
- state and action spaces
- learning algorithm
- state action
- markov decision processes
- temporal difference
- multi agent
- reinforcement learning algorithms
- state space
- partial observability
- situation calculus
- decision theoretic
- macro actions
- multiagent reinforcement learning
- learning process
- decision making
- temporal difference learning
- dynamic programming
- learning agent
- plan recognition
- function approximators
- previously learned
- continuous state
- initial state
- agent learns
- search space
- supervised learning
- model free
- robotic control
- machine learning