---
title: "校准通道决定贝叶斯错误代理：温度诱导失真的精确律"
source: "arXiv AI/CL/LG（论文）"
source_tier: "t2"
url: "https://arxiv.org/abs/2607.18162v1"
permalink: "https://feed.astralor.com/items/itm_458a85b6440e90a75a3acfd3"
published: "2026-07-20T17:07:26.000Z"
score: 76
channel: "ai"
tags: ["arXiv","校准","温度缩放","贝叶斯错误代理"]
---

# 校准通道决定贝叶斯错误代理：温度诱导失真的精确律

## AI 点评

> 这篇论文的数学工具非常漂亮，直接把温度缩放对错误代理的影响写成了一个用边际分布表达的恒等式，让原本看似随意的扭曲变得完全可预测。如果你在做模型校准或者可靠性评估，特别是用soft-label方法时，这篇文章值得细读：它提醒你，任何代理值都必须放在校准方法上下文中理解，否则很容易被误导。对研究贝叶斯错误估计和校准理论的人尤其重要。

## 速览

这篇论文精确刻画了温度缩放（最常用的后验校准方法）如何扭曲软标签贝叶斯错误代理。作者证明了一个无模型的恒等式，将温度缩放的代理简化为分类器的边际分布，从而得到温度严格单调性及从温度轴到(0,1/2)的连续双射，意味着固定分类器可报告任意代理值。在CIFAR-10等数据集上，代理值在恒定测试误差下变化56倍至980倍，且最小化期望校准误差的温度不与任何稳定代理值重合。结果强化了代理值只有与产生其概率的机制一起才有意义的建议。

## 关键点

- 证明温度缩放后的贝叶斯错误代理严格单调于温度，且温度轴与(0,1/2)之间存在连续双射
- 基于高斯logit模型推导出代理-温度曲线的封闭形式（两个参数）
- 在八个二分类任务中，代理值在恒定测试误差下变化56倍至980倍
- 最小化期望校准误差的温度不与任何稳定代理值重合

## 全文

The soft-label Bayes-error estimator beta(z) = E[min(z, 1-z)] of Ishida et al. estimates the irreducible error of a binary task directly from probability-valued labels. Recent work by Ushio et al. showed that this estimator is fragile when the probabilities are not the true posterior: even perfectly calibrated soft labels can yield a substantially inaccurate estimate, and they propose isotonic calibration as a consistent remedy. We complement that line of work by characterizing exactly how the most widely used post-hoc calibration map -- temperature scaling -- distorts the proxy. We prove an exact, model-free identity reducing the temperature-scaled proxy to the classifier's margin distribution, from which we obtain (i) strict monotonicity in the temperature and (ii) a continuous bijection from the temperature axis onto the open interval (0, 1/2), so that a fixed classifier -- with fixed decisions and fixed 0-1 error -- can be made to report any proxy value whatsoever. Under a Gaussian model of the logits we further derive a two-parameter closed form for the entire proxy-versus-temperature curve. Across CIFAR-10, Fashion-MNIST, and SVHN (eight binary tasks), the proxy varies by 56x to 980x at constant test error, the closed form reproduces the empirical curve to within 0.018, and the calibration temperature that minimizes the expected calibration error does not coincide with any stable proxy value. Our results give a precise, predictive account of the distortion whose existence motivates calibration-based remedies, and they reinforce the practical recommendation that a proxy value is meaningful only together with the mechanism that produced its probabilities.

---

引用前请核对原文;本文由星际流动整理:https://feed.astralor.com/items/itm_458a85b6440e90a75a3acfd3