Human Feedback as Action Assignment in Interactive Reinforcement Learning.
Syed Ali RazaMary-Anne WilliamsPublished in: ACM Trans. Auton. Adapt. Syst. (2020)
Keyphrases
- reinforcement learning
- sensory inputs
- action selection
- reward signal
- partially observable domains
- human subjects
- guided exploration
- human interaction
- user friendly
- machine learning
- human users
- markov decision processes
- virtual reality
- function approximation
- human actions
- computer graphics
- optimal policy
- human operators
- reinforcement learning algorithms
- virtual humans
- partially observable
- initial state
- action space
- state action
- user interaction
- supervised learning
- state space