研究人员开发了一种名为期望一致性损失（ECL）的新方法，用于提高分类模型在处理协变量偏移时的置信度校准能力。该方法通过建立“期望一致性条件”来重新思考校准，该条件表明数据分布的变化并不必然导致置信度校准不准确。ECL是一种无监督域适应损失，可应用于各种校准类型，并被证明与期望校准误差（ECE）等现有方法具有相似的样本复杂度。ECL的有效性已在模拟和真实数据集上得到验证。 AI
影响 通过在数据条件变化的情况下改进置信度校准，提高了AI模型在安全关键应用中的可靠性。
排序理由 介绍机器学习新方法的学术论文。
AI 生成摘要 · Google Gemini · 来自 2 个来源。 我们如何撰写摘要 →
arXiv:2605.21552v1 Announce Type: cross Abstract: Confidence calibration for classification models is vital in safety-critical decision-making scenarios and has received extensive attention. General confidence calibration methods assume training and test data are independent and …
Confidence calibration for classification models is vital in safety-critical decision-making scenarios and has received extensive attention. General confidence calibration methods assume training and test data are independent and identically distributed, limiting their effectiven…