Exploiting Action Impact Regularity and Exogenous State Variables for Offline Reinforcement Learning (Abstract Reprint).
Vincent LiuJames R. WrightMartha WhitePublished in: AAAI (2024)
Keyphrases
- state variables
- state space
- reinforcement learning
- action space
- dynamic systems
- action selection
- reward function
- partially observable domains
- reward shaping
- heuristic search
- function approximation
- reinforcement learning algorithms
- markov decision processes
- partially observable
- causal graph
- initial state
- state action
- optimal policy
- factored markov decision processes
- dynamic bayesian networks
- random variables
- dynamic programming
- learning algorithm
- markov decision process
- temporal difference
- multi agent
- machine learning
- learning agent
- autoregressive
- dynamical systems
- transition model
- particle filter
- partial least square regression