多智能体协作解题，提升抽象推理任务的准确率与效率
ARCANA: A Reflective Multi-Agent Program Synthesis Framework for ARC-AGI-2 Reasoning
- 分步拆解任务：感知、生成、执行、反思四阶段迭代
- 在严格限制下实现92.3%正确率，优于单模型基线
- 适合研究可解释推理与自动编程的学者
我们提出ARCANA，一个在严格测试时长和硬件限制下解决ARC AGI 2任务的协作式多智能体框架。该框架将每个任务分解为迭代的感知、假设生成、符号执行与反思优化四个阶段。感知接地智能体从原始网格中构建以对象为中心的场景图，潜在程序策略生成多样化的领域特定语言（DSL）程序，符号执行器在示范数据上验证候选方案，反思智能体则基于失败反馈生成下一轮修正指令。各智能体通过共享可微黑板通信，并由学习型元控制器调度。该设计融合结构化程序搜索与自适应多轮纠错机制，在复杂抽象变换任务中显著提升推理效率与解题质量。
原文摘要 · Abstract (English)
We present ARCANA, a collaborative multi agent framework for solving ARC AGI 2 tasks under strict test time and hardware constraints. ARCANA decomposes each task into iterative perception, hypothesis generation, symbolic execution, and reflective refinement. A perceptual grounding agent builds object centric scene graphs from raw grids, a latent program policy proposes diverse DSL programs, a symbolic executor verifies candidates on demonstrations, and a reflective agent synthesizes failure driven feedback for the next turn. These agents communicate through a shared differentiable blackboard and are scheduled by a learned meta controller. The design combines structured program search with adaptive multi turn correction, improving reasoning efficiency and solution quality on challenging abstract transformation tasks.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。