Skip to main content
QUICK REVIEW

[论文解读] Geometry-Aware Uncertainty Quantification via Conformal Prediction on Manifolds

Marzieh Amiri Shahbazi, Ali Baheri|arXiv (Cornell University)|Feb 17, 2026
Morphological variations and asymmetry被引用 0
一句话总结

论文提出自适应测地距离同构预测,用于流形值回归,产生几何感知的预测区域,并在局部自适应校准以实现对流形如 S^2 和 T^2 的统一条件覆盖率。

ABSTRACT

Conformal prediction provides distribution-free coverage guaranties for regression; yet existing methods assume Euclidean output spaces and produce prediction regions that are poorly calibrated when responses lie on Riemannian manifolds. We propose \emph{adaptive geodesic conformal prediction}, a framework that replaces Euclidean residuals with geodesic nonconformity scores and normalizes them by a cross-validated difficulty estimator to handle heteroscedastic noise. The resulting prediction regions, geodesic caps on the sphere, have position-independent area and adapt their size to local prediction difficulty, yielding substantially more uniform conditional coverage than non-adaptive alternatives. In a synthetic sphere experiment with strong heteroscedasticity and a real-world geomagnetic field forecasting task derived from IGRF-14 satellite data, the adaptive method markedly reduces conditional coverage variability and raises worst-case coverage much closer to the nominal level, while coordinate-based baselines waste a large fraction of coverage area due to chart distortion.

研究动机与目标

  • 解决对流形取值响应的不确定性量化问题,因为标准的欧几里得 conformal 预测在几何和异方差性下失效。
  • 使用几何感知的非合格分数,利用测地距离。
  • 引入局部自适应难度校准,以在不同预测难度下实现覆盖率平衡。

提出的方法

  • 在流形上使用拆分 conformal 预测与一个基预测器。
  • 用测地非合规性分数替代欧几里得残差。
  • 通过交叉验证的难度估计器对残差进行归一化,以实现局部自适应。
  • 将预测区域生成为测地圆盘(盖帽),半径与 conformal 分位数及估计的难度成正比。
  • 使用训练数据的交叉验证残差来训练难度估计器,以保持交换性。
  • 提供一个基于难度估计器与实际残差相关性的诊断,用以指导方法选择。
Figure 1: Standard geodesic CP uses fixed-radius caps for every test point (left), while adaptive geodesic CP varies cap size by local difficulty (right).
Figure 1: Standard geodesic CP uses fixed-radius caps for every test point (left), while adaptive geodesic CP varies cap size by local difficulty (right).

实验结果

研究问题

  • RQ1同态地图的 intrinsic 几何结构是否允许有效地将 conformal 预测应用于流形取值响应?
  • RQ2自适应的测地非合规性分数是否比非自适应或基于坐标的方法提供更均匀的条件覆盖?
  • RQ3在具有异方差性的合成球面数据以及真实的地磁场预测数据上,该方法表现如何?
  • RQ4什么诊断可以帮助实务工作者在自适应与非自适应方法之间进行选择?

主要发现

CaseMethodMarg. Cov.Area (sr)Cond. Std ↓Worst Cov ↑
Case 1: Synthetic SphereAdaptive Geodesic0.906±0.0201.8650.0420.839
Case 1: Synthetic SphereStandard Geodesic0.904±0.0191.8850.0520.814
Case 1: Synthetic SphereNaive Coordinate0.904±0.0192.3760.0670.784
Case 2: IGRF-14 Geomagnetic ForecastingAdaptive Geodesic0.902±0.0130.0380.0310.855
Case 2: IGRF-14 Geomagnetic ForecastingStandard Geodesic0.902±0.0130.0390.1070.689
Case 2: IGRF-14 Geomagnetic ForecastingNaive Coordinate0.903±0.0150.0460.0600.805
  • 自适应测地几何 conformal 预测在不同难度水平下的条件覆盖率更均匀,相较基线方法有改进。
  • 在合成的 S^2 数据上,自适应方法减少了条件覆盖率的标准差并提升了最坏情况覆盖。
  • 在 IGRF-14 地磁预测上,自适应方法使条件覆盖标准差下降了 71%,最坏情况覆盖从 0.689 提升至 0.855。
  • 测地盖帽保持与位置无关的面积,避免了在坐标方法中因图表扭曲而膨胀覆盖区域的问题。
  • 一个诊断性相关性(难度估计器与真实残差之间的相关性)可作为实际部署的指导(在 IGRF-14 中 r=0.516)。
  • 坐标基的朴素方法由于球面的度量扭曲,较之测地方法会浪费更多的区域面积。
Figure 2: 300-trial Monte Carlo distributions. (a) Valid marginal coverage for all methods. (b) Naive coordinate wastes 26% more area. (c) Adaptive method achieves lowest conditional coverage variance. (d) Adaptive worst-case coverage stays closest to the 0.90 target.
Figure 2: 300-trial Monte Carlo distributions. (a) Valid marginal coverage for all methods. (b) Naive coordinate wastes 26% more area. (c) Adaptive method achieves lowest conditional coverage variance. (d) Adaptive worst-case coverage stays closest to the 0.90 target.

更好的研究,从现在开始

从论文设计到论文写作,大幅缩短您的研究时间。

无需绑定信用卡

本解读由 AI 生成,并经人工编辑审核。