The Terminal Representation in Reinforcement Learning

Amir Esterhuysen, Anders Jonsson

Representation learning is a powerful tool for spatio-temporal abstraction within reinforcement learning (RL). Two well established approaches are through the successor representat…