Stochastic Q-learning for Large Discrete Action Spaces

arXiv:2405.10310v1 Announce Type: new
Abstract: In complex environments with large discrete action spaces, effective decision-making is critical in reinforcement learning (RL). Despite the widespread use of value-based RL approaches like Q-learning, they come with a computational burden, necessitating the maximization of a value function over all actions in each iteration. This burden becomes particularly challenging when addressing large-scale problems and using deep neural networks as function approximators. In this paper, we present stochastic value-based RL approaches which, in each iteration, …

abstract action arxiv computational cs.ai cs.lg cs.pf cs.ro decision environments function iteration making q-learning reinforcement reinforcement learning spaces stat.ml stochastic type value