Fully Learnable Neural Reward Machines

Hazem Dewidar, Elena Umili

Non-Markovian Reinforcement Learning (RL) tasks present significant challenges, as agents must reason over entire trajectories of state-action pairs to make optimal decisions. A co…