From Self-Confidence to Statistical Accuracy: Scaling Human-in-the-Loop Evaluation with Direct and LLM-Assisted Auditing
作者: Harine Choi, Chenchen Kuai, Zihao Li, Jiwan Jiang, Yunlong Zhang, Zhiwen Fan, Zhengzhong Tu, Yang Zhou
分类: cs.CL
发布日期: 2026-10-26
备注: 21 pages, 8 figures. Reproduction artifacts available at https://github.com/harinechoi/confidence-auditing-reproduction
💡 一句话要点
提出统一统计框架以提升人机协作评估效率
🎯 匹配领域: 支柱九:具身大模型 (Embodied Foundation Models)
关键词: 大型语言模型 多模态学习 自信度校准 统计审计 人机协作
📋 核心要点
- 现有方法中,模型报告的自信度常常误校准,影响评估的准确性和效率。
- 本文提出统一的统计框架,结合信心分层抽样和LLM辅助评估,提升审计效率。
- 实验结果显示,比例分配在不平衡自信度下显著降低估计误差,验证了方法的有效性。
📝 摘要(中文)
对于大型语言模型(LLMs)和多模态大型语言模型(MLLMs),模型报告的自信度往往存在误校准,但仍可用于高效评估。本文开发了一个统一的统计框架,支持直接审计和LLM辅助审计两种评估方式。通过信心分层抽样,直接审计能够从部分验证的输出中恢复总体准确性,而无需校准的自信度。实验结果表明,在高度不平衡的自信度分布下,比例分配显著降低了估计误差,且LLM辅助评估的自动推理成本可能限制潜在节省。研究澄清了信心导向审计和自动验证的统计及经济条件。
🔬 方法详解
问题定义:本文旨在解决大型语言模型自信度误校准导致的评估效率低下问题。现有方法在直接审计和LLM辅助审计中存在统计不准确和成本高的问题。
核心思路:论文提出的核心思路是通过信心分层抽样和误差校正估计,提升审计的统计有效性和经济性。设计上考虑了自信度的分布特性,以优化审计过程。
技术框架:整体架构包括两个主要模块:直接审计模块和LLM辅助评估模块。直接审计模块使用信心分层抽样来恢复总体准确性,而LLM辅助模块则通过建模敏感性和假阳性率来进行误差校正。
关键创新:最重要的技术创新在于提出了信心分层抽样的使用,能够在不需要校准自信度的情况下进行有效的统计审计。这与传统方法的依赖于校准自信度的方式有本质区别。
关键设计:在设计中,采用了比例分配策略来优化标签分配,并通过敏感性和假阳性率的建模来提升LLM辅助评估的准确性。
📊 实验亮点
实验结果表明,在MMLU-Pro和MMMU-Pro数据集上,采用比例分配的审计方法相较于均匀分配显著降低了估计误差,提升幅度达到XX%。此外,LLM辅助评估的有效性在特定条件下得到了验证,显示出其在实际应用中的潜力。
🎯 应用场景
该研究的潜在应用领域包括大型语言模型的评估、自动化审核系统以及人机协作的智能决策支持。通过提升审计效率和准确性,能够在实际应用中降低成本,提高系统的可靠性和用户信任度。
📄 摘要(原文)
For large language models (LLMs) and multimodal large language models (MLLMs), model-reported self-confidence is often miscalibrated, yet it may still be useful for efficient evaluation. At the same time, LLM-as-a-judge offers a scalable alternative to direct human auditing, but its practical benefit depends on calibration quality and inference cost. This paper develops a unified statistical framework for both evaluation routes under alternative sampling strategies. For direct auditing, confidence-stratified sampling is used to recover population accuracy from partially verified outputs without requiring calibrated confidence. Finite-sample uncertainty is characterized, and the analysis shows how highly uneven confidence strata affect audit efficiency and label allocation. For judge-assisted evaluation, the judge is modeled through sensitivity and false-positive rate, yielding an error-corrected estimator with calibration uncertainty. The two routes are then compared through human-label and total-cost efficiency boundaries. Experiments on MMLU-Pro and MMMU-Pro show that miscalibrated self-confidence can still support statistically valid stratified auditing. Under highly imbalanced confidence distributions, proportional allocation substantially reduces estimation error relative to equal-per-stratum allocation. For judge-assisted evaluation, only a small subset of target--verifier pairs satisfies the idealized human-label efficiency condition, and automated inference costs can further limit potential savings. These findings clarify the statistical and economic conditions under which confidence-guided auditing and automated verification are useful.