A Boolean Task Algebra for Reinforcement Learning
This addresses the problem of lifelong learning and skill composition for AI agents, offering a foundational approach with broad applicability.
The paper formalizes task composition in reinforcement learning as a Boolean algebra, enabling agents to solve new tasks by logically combining learned base skills without additional learning, and demonstrates this in domains including a high-dimensional video game, solving a super-exponential number of tasks.
The ability to compose learned skills to solve new tasks is an important property of lifelong-learning agents. In this work, we formalise the logical composition of tasks as a Boolean algebra. This allows us to formulate new tasks in terms of the negation, disjunction and conjunction of a set of base tasks. We then show that by learning goal-oriented value functions and restricting the transition dynamics of the tasks, an agent can solve these new tasks with no further learning. We prove that by composing these value functions in specific ways, we immediately recover the optimal policies for all tasks expressible under the Boolean algebra. We verify our approach in two domains---including a high-dimensional video game environment requiring function approximation---where an agent first learns a set of base skills, and then composes them to solve a super-exponential number of new tasks.