The Option-Critic Architecture
This addresses the problem of scaling reinforcement learning for AI systems by enabling autonomous temporal abstraction, though it builds incrementally on existing options frameworks.
The paper tackled the challenge of autonomously learning temporal abstractions in reinforcement learning by proposing the option-critic architecture, which learns internal policies and termination conditions for options without extra rewards, achieving flexibility and efficiency in discrete and continuous environments.
Temporal abstraction is key to scaling up learning and planning in reinforcement learning. While planning with temporally extended actions is well understood, creating such abstractions autonomously from data has remained challenging. We tackle this problem in the framework of options [Sutton, Precup & Singh, 1999; Precup, 2000]. We derive policy gradient theorems for options and propose a new option-critic architecture capable of learning both the internal policies and the termination conditions of options, in tandem with the policy over options, and without the need to provide any additional rewards or subgoals. Experimental results in both discrete and continuous environments showcase the flexibility and efficiency of the framework.