Unraveling the Complexity of Memory in RL Agents: an Approach for Classification and Evaluation

Egor Cherepanov, Nikita Kachaev, Artem Zholus +2

The incorporation of memory into agents is essential for numerous tasks within the domain of Reinforcement Learning (RL). In particular, memory is paramount for tasks that require…