arXiv：cs.LG· Minxing Zheng, Shixiang Zhu·· 3 小时前
CASA：样本高效的生成式共形预测
Sample-Efficient Generative Conformal Prediction
AI 导读
CASA（Conformal Adaptive Sample Allocation）通过刻画额外样本的边际价值，在不同输入间自适应分配采样预算，在满足边际覆盖与平均采样预算约束下最小化期望集合大小。
正文
Abstract:Generative conformal prediction builds uncertainty sets from samples of a conditional generator, which are efficient only when the samples represent the response distribution well. This can require many samples, each of which can be costly, as in large diffusion models and scientific simulators, so the sampling budget must be used efficiently. Existing methods draw the same number of samples at every input, wasting samples where the response distribution is simple and undersampling where it is complex, which inflates sets and leaves those inputs under-covered. We propose CASA (Conformal Adaptive Sample Allocation), which characterizes the marginal value of an additional sample and allocates samples across inputs to minimize the expected set size subject to marginal coverage and an average sampling budget. Theoretical analysis shows that adaptive allocation yields smaller sets than a fixed count at the same budget: a missed mode forces a radius that spans the gap between modes, and even oracle radius cannot compensate for it. On synthetic and real tasks, CASA produces substantially smaller sets at the same budget, often improves conditional coverage, and complements existing radius-adaptive methods.
| Subjects: | Machine Learning (cs.LG); Methodology (stat.ME) |
| Cite as: | arXiv:2610.11349 [cs.LG] |
| (or arXiv:2610.11349v1 [cs.LG] for this version) | |
| https://doi.org/10.48550/arXiv.2610.11349 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Minxing Zheng [view email]
[v1]
Thu, 8 Oct 2026 06:44:45 UTC (4,160 KB)
来源：arXiv：cs.LG · arxiv.org