Definition
A parallelized reinforcement learning method that utilizes multiple, independent agents to interact with the environment and update a central policy and value function.
In short: A parallelized reinforcement learning method that utilizes multiple, independent agents to interact with the environment and update a central policy and value function.