Distribution Calibration for Regression

Hao Song, Tom Diethe, Meelis Kull, Peter Flach

Research output: Chapter in Book/Report/Conference proceedingConference Contribution (Conference Proceeding)

155 Downloads (Pure)

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 distribution calibration, and demonstrate its advantages over the existing definition of quantile calibration. We further propose a post-hoc approach to improving the predictions from previously trained regression models, using multi-output Gaussian Processes with a novel Beta link function. The proposed method is experimentally verified on a set of common regression models and shows improvements for both distribution-level and quantile-level calibration.
Original languageEnglish
Title of host publicationInternational Conference on Machine Learning, 9-15 June 2019, Long Beach, California, USA
EditorsKamalika Chaudhuri, Ruslan Salakhutdinov
PublisherProceedings of Machine Learning Research
Pages5897-5906
Number of pages10
Publication statusPublished - 15 May 2019

Publication series

NameProceedings of Machine Learning Research
Volume97
ISSN (Print)2640-3498

Research Groups and Themes

  • Digital Health
  • SPHERE

Keywords

  • stat.ML
  • cs.AI
  • cs.LG

Fingerprint

Dive into the research topics of 'Distribution Calibration for Regression'. Together they form a unique fingerprint.

Cite this