Action Branching Architectures for Deep Reinforcement Learning.
Arash TavakoliFabio PardoPetar KormushevPublished in: AAAI (2018)
Keyphrases
- reinforcement learning
- action selection
- partially observable domains
- action space
- reward shaping
- agent learns
- state action
- function approximation
- search tree
- learning algorithm
- fitted q iteration
- transition model
- reinforcement learning algorithms
- dynamic programming
- optimal control
- model free
- state space
- continuous state
- multi agent
- robotic control
- computer vision
- evaluation function
- branch and bound
- deep learning
- markov decision processes
- sensory inputs
- learning process
- genetic algorithm