Transfer reinforcement learning is the surging research area for improving knowledge transfer methods from a set of source tasks to the target tasks. Generally, Reinforcement learning is one of the machine learning algorithms that refers to the problem an agent faces that learns behavior through trial-and-error interactions with a dynamic environment and transfer learning algorithm automatically utilizes the prior knowledge learned from the solving relevant source tasks for the learning process of new tasks.

The combination of reinforcement and transfer learning significantly improves the performance of the system and learning efficiency of agents by trained knowledge on similar tasks from other source agents.