1 paper
Hojin Ko, Jeonggyu Huh
Most value-based and actor--critic reinforcement learning methods rely on Bellman-style recursions, yet these recursions collapse under non-exponential discounting common in human…