Robust Successor Features

Erik Nikulski, Yamen Habib, Vicenç Gomez +3

Generalization in Reinforcement Learning (RL) refers to the ability to execute close-to-optimal policies in unseen tasks after the agent has been trained on a different set of task…