BriefGPT.xyz
May, 2019
回归分布校准
Distribution Calibration for Regression
HTML
PDF
Hao Song, Tom Diethe, Meelis Kull, Peter Flach
TL;DR
本文介绍了一种新的分布校准方法,并提出了采用多输出高斯过程和Beta链接函数的方法来改善先前训练的回归模型的预测的后验方法。该方法得到了实验验证,并在分布层和分位数层面上均有所提高。
Abstract
We are concerned with obtaining well-calibrated output distributions from
regression models
. Such distributions allow us to quantify the uncertainty that the model has regarding the predicted target value. We introduce the novel concept of
→