发表机构
University of Illinois Urbana-Champaign; Harvard University; University of Chicago（伊利诺伊大学厄巴纳-香槟分校; 哈佛大学; 芝加哥大学）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
本文提出滚动共形预测（rolling-CP），一种无需数据分割的序贯模型训练推断方法，在可交换数据下提供边际覆盖保证，并在实验中验证其有效性。
AI 中文摘要
我们提出了滚动共形预测（rolling-CP），一种适用于序贯模型训练场景的无分布预测推断方法。具体而言，给定数据流$(X_1,Y_1),(X_2,Y_2),\dots$，在每次时间$n$，训练好的模型可能依赖于观测到的历史数据$\{(X_i,Y_i)\}_{i<n}$。这种设置自然出现在现代序贯训练中，包括对大规模数据集的一次性训练，以及语言模型在部署期间的持续微调或测试时自适应。滚动共形预测首先将每个新观测值针对当前预测器进行校准，然后将其滚动到未来的训练中。通过这种方式，我们避免了数据分割的需求。值得注意的是，尽管时间$n=1,2,\dots$处的模型可能具有完全不同的性质和准确度水平，但对于可交换数据，仍然可以建立边际覆盖保证，具有熟悉的通用因子二保证（最坏情况下的$1-2α$覆盖，相对于目标水平$1-α$），且无需任何稳定性假设或对模型训练过程的任何限制。对于独立同分布（i.i.d.）数据流，我们进一步证明了随时间均匀的高概率训练条件有效性；在稳定性条件下，覆盖保证可加强至接近$1-α$。在序贯回归、多类随机梯度下降（SGD）和一次性神经网络训练上的数值实验进一步证明了滚动共形预测的实际有效性。
英文摘要
We introduce Rolling Conformal Prediction (rolling-CP), a distribution-free predictive inference method for the setting of sequential model training. Specifically, given a data stream $(X_1,Y_1),(X_2,Y_2),\dots$, at each time $n$ the trained model may depend on the observed history $\{(X_i,Y_i)\}_{i<n}$. This setting arises naturally in modern sequential training, including one-pass training over massive datasets and continual fine-tuning or test-time adaptation of language models during deployment. Rolling-CP first calibrates each incoming observation against the current predictor and then rolls it into future training. In this way, we avoid the need for data splitting. Remarkably, although the models at times $n=1,2,\dots$ may have entirely different properties and accuracy levels, for exchangeable data it is nonetheless possible to establish a guarantee of marginal coverage, with a familiar universal factor-two guarantee (a worst case guarantee of $1-2α$ coverage, as compared to the target level $1-α$), without any assumptions of stability or any restrictions on the model training process. For i.i.d. data streams, we further prove high-probability training-conditional validity uniformly over time; under stability conditions, coverage guarantees sharpen towards $1-α$. Numerical experiments on sequential regression, multiclass SGD, and one-pass neural-network training further demonstrate the practical effectiveness of rolling-CP.
Comments42 pages, 5 figures
