用玻尔兹曼影响函数测量随机数据复杂度
Measuring Stochastic Data Complexity with Boltzmann Influence Functions
AI总结：
提出IF-COMP，用温度标度玻尔兹曼影响函数线性化模型以近似pNML分布，实现不确定性校准与数据复杂度测量，在多项任务中匹配或超越强基线。
摘要：
估计模型在测试点上的预测不确定性是确保分布偏移下可靠性和校准的关键部分。针对这一问题，最小描述长度方法使用预测归一化最大似然（pNML）分布，该分布考虑数据点的每一个可能标签，如果其他标签也与模型和训练数据一致，则降低预测的置信度。在这项工作中，我们提出了IF-COMP，一种可扩展且高效的pNML分布近似方法，它通过温度标度的玻尔兹曼影响函数对模型进行线性化。IF-COMP可用于在测试点上生成良好校准的预测，以及在标记和未标记设置中测量复杂度。我们在不确定性校准、错误标签检测和OOD检测任务上对IF-COMP进行了实验验证，在这些任务中，它始终匹配或超越强基线方法。
英文摘要：
Estimating the uncertainty of a model's prediction on a test point is a crucial part of ensuring reliability and calibration under distribution shifts. A minimum description length approach to this problem uses the predictive normalized maximum likelihood (pNML) distribution, which considers every possible label for a data point, and decreases confidence in a prediction if other labels are also consistent with the model and training data. In this work we propose IF-COMP, a scalable and efficient approximation of the pNML distribution that linearizes the model with a temperature-scaled Boltzmann influence function. IF-COMP can be used to produce well-calibrated predictions on test points as well as measure complexity in both labelled and unlabelled settings. We experimentally validate IF-COMP on uncertainty calibration, mislabel detection, and OOD detection tasks, where it consistently matches or beats strong baseline methods.
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。
