TreeQN and ATreeC: Differentiable Tree Planning for Deep Reinforcement Learning.
Gregory FarquharTim RocktäschelMaximilian IglShimon WhitesonPublished in: CoRR (2017)
Keyphrases
- reinforcement learning
- action selection
- tree structure
- objective function
- function approximation
- partially observable
- deterministic domains
- macro actions
- planning problems
- heuristic search
- state space
- motion planning
- model free
- ai planning
- optimal policy
- stochastic domains
- data structure
- partially observable domains
- optimal control
- r tree
- multi agent
- spanning tree
- hierarchical structure
- tree structures
- supervised learning
- deep learning
- planning process
- markov decision problems
- learning process
- learning algorithm
- machine learning