Numeric Reward Machines

arXiv:2404.19370v1 Announce Type: cross
Abstract: Reward machines inform reinforcement learning agents about the reward structure of the environment and often drastically speed up the learning process. However, reward machines only accept Boolean features such as robot-reached-gold. Consequently, many inherently numeric tasks cannot profit from the guidance offered by reward machines. To address this gap, we aim to extend reward machines with numeric features such as distance-to-gold. For this, we present two types of reward machines: numeric-Boolean and numeric. In a …

abstract agents aim arxiv cs.ai cs.lg environment features gap guidance however machines process profit reinforcement reinforcement learning robot speed tasks the environment type