arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2603.27414math.STcs.AIstat.TH
收藏

多预测驱动推断

Multiple-Prediction-Powered Inference

Massachusetts Institute of Technology(麻省理工学院);Google Research(谷歌研究院);Google DeepMind(谷歌DeepMind)

Charlie Cowen-Breen, Alekh Agarwal, Stephen Bates, William W. Cohen, Jacob Eisenstein, Amir Globerson, Adam Fisch

AI总结:

本文提出多预测驱动推断框架,通过优化分配资源于不同数据源,实现统计高效估计。理论证明其最小最大最优性、有限样本性能和渐近正态性,并在三个大语言模型评估场景中验证了其优于现有基线的估计误差。

摘要:

统计估计常需在昂贵高质量测量与多种低质量代理之间进行权衡。我们引入多预测驱动推断(MultiPPI):一种通用框架,通过优化分配资源于这些多样化数据源来构建统计高效估计。本文提供了关于MultiPPI估计器最小最大最优性、有限样本性能和渐近正态性的理论保证。通过在三个多样化的大型语言模型(LLM)评估场景中的实验,我们证明MultiPPI始终比现有基线具有更低的估计误差。这种优势源于其预算自适应分配策略,该策略通过学习模型的复杂成本和相关结构,战略性地组合模型子集。

英文摘要:

Statistical estimation often involves tradeoffs between expensive, high-quality measurements and a variety of lower-quality proxies. We introduce Multiple-Prediction-Powered Inference (MultiPPI): a general framework for constructing statistically efficient estimates by optimally allocating resources across these diverse data sources. This work provides theoretical guarantees about the minimax optimality, finite-sample performance, and asymptotic normality of the MultiPPI estimator. Through experiments across three diverse large language model (LLM) evaluation scenarios, we show that MultiPPI consistently achieves lower estimation error than existing baselines. This advantage stems from its budget-adaptive allocation strategy, which strategically combines subsets of models by learning their complex cost and correlation structures.

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。

↑