Causal versus Marginal Shapley Values for Robotic Lever Manipulation Controlled using Deep Reinforcement Learning.
Sindre Benjamin RemmanInga StrümkeAnastasios M. LekkasPublished in: ACC (2022)
Keyphrases
- reinforcement learning
- manipulation tasks
- robotic systems
- probability distribution
- parameter values
- mobile robot
- dynamic programming
- real robot
- marginal distributions
- game theory
- robot control
- function approximation
- conditional probabilities
- real time
- optimal policy
- learning algorithm
- machine learning
- markov decision processes
- random variables
- causal models
- causal relationships
- model free
- state space