Adaptive Conformal Inference Under Distribution Shift We test our method, adaptive conformal inference, on two real world datasets and find that its predictions are robust to visible and significant distribution shifts
Adaptive Conformal Inference Under Distribution Shift We test our method, adaptive conformal inference, on two real world datasets and find that its predictions are robust to visible and significant distribution shifts
Adaptive conformal inference under distribution shift Adaptive Conformal Inference (ACI) [Gibbs et al , 2021] is a method that aims at correcting the coverage of conformal prediction methods in a sequential prediction framework (e g time series forecasting) when the distribution of the data shifts over time
Adaptive Conformal Inference Under Distribution Shift Oral Adaptive Conformal Inference Under Distribution Shift Isaac Gibbs · Emmanuel Candes [ Abstract ] [ Visit Oral Session 5: Optimization and Vision Applications ]
Adaptive Conformal Inference Under Distribution Shift We test our method, adaptive conformal inference, on two real world datasets and find that its predictions are robust to visible and significant distribution shifts
分布偏移与保形推断 - Tan Jay | 唐 洁 在机器学习中, Distribution Shift(分布偏移) 和 Conformal Inference(保形推断) 的关系可以从以下角度分析: 1 核心挑战:可交换性假设的违背 保形推断依赖数据的 可交换性(exchangeability),即数据的顺序不影响联合分布。 这一假设在独立同分布(i i d )或有限总体不放回抽样时成立。 当训练数据与测试数据分布不一致(如协变量偏移、标签偏移等),可交换性假设被打破,传统保形推断的覆盖概率保证(如 95% 置信水平)可能失效。 2 覆盖概率的退化 传统保形推断保证 整体数据集 的覆盖概率,但无法约束 特定子群体或条件 下的覆盖。 示例:在医疗数据中,模型对多数群体的覆盖概率为 95%,但对少数群体可能降至 80%。
Adaptive Conformal Inference Under Distribution Shift A detailed description of five Adaptive Conformal Inference algorithms and their theoretical guarantees are provided, and a case study of producing prediction intervals for influenza incidence in the United States based on black-box point forecasts is presented
Adaptive Conformal Inference Under Distribution Shift We test our method, adaptive conformal inference, on two real world datasets and find that its predictions are robust to visible and significant distribution shifts