Conformalized Kernel ridge regression

Evgeny Burnaev, Ivan Nazarov

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

20 Citations (Scopus)

Abstract

General predictive models do not provide a measure of confidence in predictions without Bayesian assumptions. A way to circumvent potential restrictions is to use conformal methods for constructing non-parametric confidence regions, that offer guarantees regarding validity. In this paper we provide a detailed description of a computationally efficient conformal procedure for Kernel Ridge Regression (KRR), and conduct a comparative numerical study to see how well conformal regions perform against the Bayesian confidence sets. The results suggest that conformalized KRR can yield predictive confidence regions with specified coverage rate, which is essential in constructing anomaly detection systems based on predictive models.

Original languageEnglish
Title of host publicationProceedings - 2016 15th IEEE International Conference on Machine Learning and Applications, ICMLA 2016
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages45-52
Number of pages8
ISBN (Electronic)9781509061662
DOIs
Publication statusPublished - 31 Jan 2017
Externally publishedYes
Event15th IEEE International Conference on Machine Learning and Applications, ICMLA 2016 - Anaheim, United States
Duration: 18 Dec 201620 Dec 2016

Publication series

NameProceedings - 2016 15th IEEE International Conference on Machine Learning and Applications, ICMLA 2016

Conference

Conference15th IEEE International Conference on Machine Learning and Applications, ICMLA 2016
Country/TerritoryUnited States
CityAnaheim
Period18/12/1620/12/16

Keywords

  • Confidence region
  • Conformal prediction
  • Gaussian process regression
  • Kernel Ridge Regression

Fingerprint

Dive into the research topics of 'Conformalized Kernel ridge regression'. Together they form a unique fingerprint.

Cite this