Online Reinforcement Learning
A reinforcement learning paradigm in which the agent continuously interacts with the environment during training, collecting data while updating its policy. It can explore states not covered by offline data, but interaction costs are high and safety risks exist.
Timeline (last 90 days)
On-policy RL requires the agent to continuously interact with the environment during training, and the interaction cost grows linearly with training scale.
In online reinforcement learning, data cannot be reused across multiple rounds, so it is extremely sensitive to sampling efficiency and throughput.
All Facts (2)
On-policy RL requires the agent to continuously interact with the environment during training, and the interaction cost grows linearly with training scale.
50%UnverifiedIn online reinforcement learning, data cannot be reused across multiple rounds, so it is extremely sensitive to sampling efficiency and throughput.
50%