Action Guidance with MCTS for Deep Reinforcement Learning.
Bilal KartalPablo Hernandez-LealMatthew E. TaylorPublished in: AIIDE (2019)
Keyphrases
- reinforcement learning
- action selection
- action space
- partially observable domains
- reward shaping
- function approximation
- monte carlo tree search
- transition model
- state action
- state space
- markov decision processes
- learning problems
- model free
- temporal difference
- robotic control
- learning algorithm
- fitted q iteration
- agent learns
- machine learning
- multi agent reinforcement learning
- monte carlo
- learning process
- sensory inputs
- action models
- partially observable
- state and action spaces
- optimal control
- optimal policy