时间化置信度：基于信号时序逻辑的思维链推理评估
Temporalizing Confidence: Evaluation of Chain-of-Thought Reasoning with Signal Temporal Logic
AI总结：
本文提出用信号时序逻辑评估思维链推理中的逐步置信度，并通过不确定性重塑策略改善LLM数学推理的校准与可靠性。
摘要：
大型语言模型（LLM）在思维链（CoT）提示的引导下，已在数学推理任务中展现出令人瞩目的性能。然而，它们往往会产生高度自信但错误的输出，这在教育等领域构成重大风险，因为这些领域的用户可能缺乏评估推理步骤的专业知识。为解决这一问题，我们提出一个结构化框架，将逐步置信度建模为时间信号，并使用信号时序逻辑（STL）对其进行评估。具体而言，我们定义了基于STL的形式化约束，以捕捉理想的时间属性，并计算鲁棒性分数，作为结构化、可解释的置信度估计。我们的方法还引入了一组不确定性重塑策略，以在整个推理轨迹中强制实现平滑性、单调性和因果一致性。实验表明，与传统的置信度聚合和事后校准相比，我们的方法持续改善了校准指标，并提供了更可靠的不确定性估计。
英文摘要：
Large Language Models (LLMs) have shown impressive performance in mathematical reasoning tasks when guided by Chain-of-Thought (CoT) prompting. However, they tend to produce highly confident yet incorrect outputs, which poses significant risks in domains like education, where users may lack the expertise to assess reasoning steps. To address this, we propose a structured framework that models stepwise confidence as a temporal signal and evaluates it using Signal Temporal Logic (STL). In particular, we define formal STL-based constraints to capture desirable temporal properties and compute robustness scores that serve as structured, interpretable confidence estimates. Our approach also introduces a set of uncertainty reshaping strategies to enforce smoothness, monotonicity, and causal consistency across the reasoning trajectory. Experiments show that our approach consistently improves calibration metrics and provides more reliable uncertainty estimates than conventional confidence aggregation and post-hoc calibration.
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。
