Bayer, Caleidgh Grace (2023-08-25)
Visual formulations of reinforcement learning tasks are potentially challenging because (1) the state space is large and composed from pixels (so unlikely to be directly correlated with actions), (2) the underlying task ...