Konidaris George
Computer Science Department, Brown University 115 Waterman Street Providence RI 02906.
Curr Opin Behav Sci. 2019 Oct;29:1-7. doi: 10.1016/j.cobeha.2018.11.005. Epub 2018 Dec 14.
A generally intelligent agent faces a dilemma: it requires a complex sensorimotor space to be capable of solving a wide range of problems, but many tasks are only feasible given the right problem-specific formulation. I argue that a necessary but understudied requirement for general intelligence is the ability to form task-specific abstract representations. I show that the reinforcement learning paradigm structures this question into how to learn action abstractions and how to learn state abstractions, and discuss the field's progress on these topics.
它需要一个复杂的感觉运动空间才能有能力解决广泛的问题，但许多任务只有在正确的特定问题表述下才可行。我认为，对于一般智能来说，一个必要但却未得到充分研究的要求是形成特定任务抽象表征的能力。我表明，强化学习范式将这个问题构建为如何学习动作抽象以及如何学习状态抽象，并讨论了该领域在这些主题上的进展。
