发表机构
Carthago Labs; IÉSEG（迦太基实验室; IESEG商学院）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
本文提出基于配对e过程的语言级路由器，在统计控制下动态选择是否翻译多语言文档，提升分类准确率并保证决策可审计。
AI 中文摘要
何时翻译多语言文档是文本分类中的一个核心路由问题：翻译可以改善某些语言的预测，同时可能损害其他语言的性能或增加不必要的计算开销。统一的翻译策略和基于启发式的语言分层无法提供具有统计控制的路由选择。我们提出了一种基于配对e过程的语言级路由器，该路由器在冻结路由策略之前，持续比较直接分类与翻译辅助分类的性能。通过家族误差控制阈值280，我们将每个数据集中14种合格语言中任何错误路由的概率限制在0.05以内。在SIB-200和MASSIVE数据集上，路由器分别为15种语言中的4种和15种语言环境中的14种选择了翻译，相比直接分类，在保留测试集上的准确率分别提升了8.14和16.70个百分点。所有28个决策在50种与结果无关的排序中以及相对于每组阈值均保持稳定。我们的结果表明，配对e过程能够实现具有统计控制、随时有效且可审计的多语言分类路由。
英文摘要
Choosing when to translate multilingual documents is a central routing problem in text classification: translation can improve predictions for some languages while degrading others or adding unnecessary computation. Uniform translation and heuristic language tiers do not provide statistically controlled route selection. We introduce a language-level router based on paired e-processes that continuously compares direct and translation-assisted classification before freezing a routing policy. A familywise-controlled threshold of 280 bounds the probability of any false route across 14 eligible languages per dataset by 0.05. On SIB-200 and MASSIVE, the router selects translation for 4 of 15 languages and 14 of 15 locales, improving held-out accuracy over direct classification by 8.14 and 16.70 percentage points, respectively. All 28 decisions remain stable across 50 outcome-independent orderings and relative to the per-group threshold. Our results demonstrate that paired e-processes enable statistically controlled, anytime-valid, and auditable multilingual classification routing.
Comments10 pages, including references and supplementary material. Workshop paper. Code: NeurIPS2026_E-Values" target="_blank" rel="noopener">https://github.com/WajdiBenSaad/NeurIPS2026_E-Values
