arXiv 2409.172702024-10-23cs.AIcs.CLcs.LGcs.LOcs.NE

思维证明：神经符号程序合成实现鲁棒且可解释的推理

Proof of Thought : Neurosymbolic Program Synthesis allows Robust and Interpretable Reasoning

Case Western Reserve University（凯斯西储大学）；Microsoft Corporation（微软公司）

Debargha Ganguly, Srinivasan Iyengar, Vipin Chaudhary, Shivkumar Kalyanaraman

备注：: 38th Conference on Neural Information Processing Systems (NeurIPS 2024) System 2 Reasoning At Scale Workshop

阅读 PDF

arXiv 页面 ↗HTML 原文 ↗

AI总结：

本文提出 Proof of Thought 框架，通过神经符号程序合成将 LLM 输出转化为一阶逻辑并由定理证明器验证，结合 JSON 领域特定语言实现鲁棒且可解释的推理，在 StrategyQA 和多模态任务上验证了有效性。

摘要：

大型语言模型（LLMs）已经彻底改变了自然语言处理，但它们在推理一致性方面仍存在困难，尤其是在新领域和复杂逻辑序列中。本研究提出了 Proof of Thought 框架，旨在增强 LLM 输出的可靠性和透明度。我们的方法将 LLM 生成的想法与形式逻辑验证相连接，采用自定义解释器将 LLM 输出转换为一阶逻辑构造，以供定理证明器审查。我们方法的核心是一种基于 JSON 的领域特定语言（Domain-Specific Language）中间表示，该语言在设计上平衡了精确的逻辑结构与直观的人类概念。这种混合表示既支持对 LLM 推理过程进行严格验证，也支持人类对其进行可理解的理解。主要贡献包括：一个具有类别（sort）管理功能的健壮类型系统，用于增强逻辑完整性；对规则的显式表示，以清晰区分事实性知识与推理性知识；以及一个灵活的架构，便于扩展到各种领域特定应用。我们通过在 StrategyQA 和一个新颖的多模态推理任务上进行基准测试，展示了 Proof of Thought 的有效性，结果表明在开放式场景中性能有所提升。通过提供可验证且可解释的结果，我们的技术解决了 AI 系统问责制的关键需求，并为高风险领域中的人类参与监督（human-in-the-loop oversight）奠定了基础。

英文摘要：

Large Language Models (LLMs) have revolutionized natural language processing, yet they struggle with inconsistent reasoning, particularly in novel domains and complex logical sequences. This research introduces Proof of Thought, a framework that enhances the reliability and transparency of LLM outputs. Our approach bridges LLM-generated ideas with formal logic verification, employing a custom interpreter to convert LLM outputs into First Order Logic constructs for theorem prover scrutiny. Central to our method is an intermediary JSON-based Domain-Specific Language, which by design balances precise logical structures with intuitive human concepts. This hybrid representation enables both rigorous validation and accessible human comprehension of LLM reasoning processes. Key contributions include a robust type system with sort management for enhanced logical integrity, explicit representation of rules for clear distinction between factual and inferential knowledge, and a flexible architecture that allows for easy extension to various domain-specific applications. We demonstrate Proof of Thought's effectiveness through benchmarking on StrategyQA and a novel multimodal reasoning task, showing improved performance in open-ended scenarios. By providing verifiable and interpretable results, our technique addresses critical needs for AI system accountability and sets a foundation for human-in-the-loop oversight in high-stakes domains.

机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。