Explainable Reinforcement Learning Based on Q-Value Decomposition by Expected State Transitions.
Yuta TsuchiyaYasuhide MoriMasashi EgiPublished in: AAAI Spring Symposium: MAKE (2023)
Keyphrases
- state transitions
- reinforcement learning
- state transition
- state action
- function approximation
- state space
- image decomposition
- markov decision processes
- learning algorithm
- reinforcement learning algorithms
- robotic control
- input output
- supervised learning
- multi agent
- optimal policy
- probability distribution
- search space
- decomposition methods
- multi agent systems
- user actions
- databases