trustworthy-ml

Tag

Cards List
#trustworthy-ml

The Hidden Costs of 99% Accuracy: A Trustworthiness Audit of the Telco Customer Churn Benchmark

arXiv cs.LG ↗ · 2026-10-03 Cached

这篇论文对IBM Telco客户流失基准进行了可信度审计,揭示了报告高达99%准确率背后的四大问题:SMOTE前划分造成的数据泄漏、TotalCharges字段冗余性导致的SHAP解释失真、概率校准方法选择,以及F1最优阈值与成本最优阈值之间的巨大差距,并提出了一个可复现的报告检查清单。

0 favorites 0 likes
#trustworthy-ml

Inducing Reward-Free Judging Rubrics that Reduce Over-Crediting in Agent Evaluation

arXiv cs.AI ↗ · 2026-08-17 Cached

RubricForge induces evaluation rubrics from labeled trajectories to reduce over-crediting in language model agent evaluation, enhancing faithfulness without environment access.

0 favorites 0 likes
← Back to home

Submit Feedback