arXiv 2210.030672022-10-06stat.MLcs.LGeess.SP

基于元学习交叉验证的共形预测集合预测量的Few-Shot校正

Few-Shot Calibration of Set Predictors via Meta-Learned Cross-Validation-Based Conformal Prediction

Sangwoo Park, Kfir M. Cohen, Osvaldo Simeone

备注：: submitted for journal publication

摘要：

已知传统的频率主义学习会产生校准不良的模型，这些模型无法可靠地量化其决策的不确定性。贝叶斯学习可以改善校准，但形式保证仅适用于对正确的模型规范的限制性假设。共形预测（CP）提供了一个通用框架，用于设计具有校准保证的集合预测器，无论底层数据生成机制如何。然而，当训练数据有限时，CP倾向于产生大的，因此没有信息的预测集。本文介绍了一种新的元学习解决方案，旨在减少集合预测大小。与先前的工作不同，所提出的元学习方案（称为元XB）（i）建立在基于交叉验证的CP上，而不是效率较低的基于验证的CP;（ii）保留正式的每任务校准保证，而不是不太严格的任务边际保证。最后，元XB扩展到自适应非共形分数，这是经验，以进一步提高边际每输入校准。

英文摘要：

Conventional frequentist learning is known to yield poorly calibrated models that fail to reliably quantify the uncertainty of their decisions. Bayesian learning can improve calibration, but formal guarantees apply only under restrictive assumptions about correct model specification. Conformal prediction (CP) offers a general framework for the design of set predictors with calibration guarantees that hold regardless of the underlying data generation mechanism. However, when training data are limited, CP tends to produce large, and hence uninformative, predicted sets. This paper introduces a novel meta-learning solution that aims at reducing the set prediction size. Unlike prior work, the proposed meta-learning scheme, referred to as meta-XB, (i) builds on cross-validation-based CP, rather than the less efficient validation-based CP; and (ii) preserves formal per-task calibration guarantees, rather than less stringent task-marginal guarantees. Finally, meta-XB is extended to adaptive non-conformal scores, which are shown empirically to further enhance marginal per-input calibration.

机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。

摘要：

英文摘要：

相关论文