Binwise Variance Scaling (BVS) has recently been proposed as a post hoc recalibration method for prediction uncertainties of machine learning regression problems that is able of more efficient corrections than uniform variance (or temperature) scaling. The original version of BVS uses uncertainty-based binning, which is aimed to improve calibration conditionally on uncertainty, i.e. consistency. I explore here several adaptations of BVS, in particular with alternative loss functions and a binning scheme based on an input-feature (X) in order to improve adaptivity, i.e. calibration conditional on X. The performances of BVS and its proposed variants are tested on a benchmark dataset for the prediction of atomization energies and compared to the results of isotonic regression.

翻译：分箱方差缩放（BVS）最近被提出作为一种针对机器学习回归问题预测不确定度的后验重校准方法，其校正效率优于均匀方差（或温度）缩放。原始BVS采用基于不确定度的分箱策略，旨在改善条件于不确定度的校准效果（即一致性）。本文探索了BVS的多种改进方案，特别是采用替代损失函数以及基于输入特征（X）的分箱方案，以提升条件于X的校准效果（即自适应性）。通过预测原子化能的基准数据集，测试了BVS及其改进变体的性能，并与等渗回归结果进行了比较。