Mixed-Policy Asynchronous Deep Q-Learning.
David SimõesNuno LauLuís Paulo ReisPublished in: ROBOT (2) (2017)
Keyphrases
- optimal policy
- action selection
- reinforcement learning
- state space
- multi agent
- cooperative
- function approximation
- reinforcement learning algorithms
- policy iteration
- markov decision processes
- reinforcement learning problems
- continuous state spaces
- learning algorithm
- deep learning
- state action
- decision problems
- stochastic approximation
- dynamic programming
- markov decision process
- discounted reward
- control policies
- approximate policy iteration
- asymptotically optimal
- neural network
- average reward
- discussion forums
- temporal difference learning
- single agent
- markov decision problems
- policy gradient
- long run
- asynchronous communication
- finite state
- sufficient conditions
- search algorithm
- agent receives