Model-Based Offline Planning

Arthur Argenson, Gabriel Dulac-Arnold

Offline learning is a key part of making reinforcement learning (RL) useable in real systems. Offline RL looks at scenarios where there is data from a system's operation, but no di…