Subgoal Search For Complex Reasoning Tasks
It addresses complex reasoning problems like puzzles and inequality proving, offering a novel approach inspired by human problem-solving, but is incremental as it builds on existing search frameworks.
The paper tackles complex reasoning tasks by proposing Subgoal Search (kSubS), a method that uses a learned subgoal generator to produce diverse, achievable subgoals, reducing search space and enabling efficient planning. It achieves state-of-the-art results on the INT benchmark and strong performance on Sokoban and Rubik's Cube within modest computational budgets.
Humans excel in solving complex reasoning tasks through a mental process of moving from one idea to a related one. Inspired by this, we propose Subgoal Search (kSubS) method. Its key component is a learned subgoal generator that produces a diversity of subgoals that are both achievable and closer to the solution. Using subgoals reduces the search space and induces a high-level search graph suitable for efficient planning. In this paper, we implement kSubS using a transformer-based subgoal module coupled with the classical best-first search framework. We show that a simple approach of generating $k$-th step ahead subgoals is surprisingly efficient on three challenging domains: two popular puzzle games, Sokoban and the Rubik's Cube, and an inequality proving benchmark INT. kSubS achieves strong results including state-of-the-art on INT within a modest computational budget.