Princeton Neuroscience Institute & Psychology Department, Princeton University, USA.
Curr Opin Neurobiol. 2010 Apr;20(2):251-6. doi: 10.1016/j.conb.2010.02.008. Epub 2010 Mar 11.
Reinforcement learning (RL) algorithms provide powerful explanations for simple learning and decision-making behaviors and the functions of their underlying neural substrates. Unfortunately, in real-world situations that involve many stimuli and actions, these algorithms learn pitifully slowly, exposing their inferiority in comparison to animal and human learning. Here we suggest that one reason for this discrepancy is that humans and animals take advantage of structure that is inherent in real-world tasks to simplify the learning problem. We survey an emerging literature on 'structure learning'--using experience to infer the structure of a task--and how this can be of service to RL, with an emphasis on structure in perception and action.
强化学习 (RL) 算法为简单的学习和决策行为及其潜在神经基质的功能提供了强大的解释。不幸的是，在涉及许多刺激和动作的现实情况下，这些算法的学习速度非常缓慢，这暴露了它们与动物和人类学习相比的劣势。在这里，我们认为造成这种差异的一个原因是人类和动物利用了现实任务中固有的结构来简化学习问题。我们调查了关于“结构学习”的新兴文献——使用经验来推断任务的结构——以及这如何对 RL 有帮助，重点是感知和行动中的结构。
