On the Challenges of Using Reinforcement Learning in Precision Drug Dosing: Delay and Prolongedness of Action Effects.
Sumana BasuMarc-André LegaultAdriana Romero-SorianoDoina PrecupPublished in: AAAI (2023)
Keyphrases
- reinforcement learning
- action selection
- indirect effects
- function approximation
- state action
- lessons learned
- action space
- reward shaping
- high precision
- temporal difference
- partially observable domains
- open issues
- key issues
- model free
- real world
- policy search
- technical challenges
- optimal control
- state space
- learning process
- neural network
- reinforcement learning algorithms
- human actions
- sensory inputs
- transition model
- learning algorithm
- genetic algorithm
- fitted q iteration