English

Conformal e-prediction

Machine Learning 2025-05-20 v5 Machine Learning

Abstract

This paper discusses a counterpart of conformal prediction for e-values, conformal e-prediction. Conformal e-prediction is conceptually simpler and had been developed in the 1990s as a precursor of conformal prediction. When conformal prediction emerged as result of replacing e-values by p-values, it seemed to have important advantages over conformal e-prediction without obvious disadvantages. This paper re-examines relations between conformal prediction and conformal e-prediction systematically from a modern perspective. Conformal e-prediction has advantages of its own, such as the ease of designing conditional conformal e-predictors and the guaranteed validity of cross-conformal e-predictors (whereas for cross-conformal predictors validity is only an empirical fact and can be broken with excessive randomization). Even where conformal prediction has clear advantages, conformal e-prediction can often emulate those advantages, more or less successfully.

Keywords

Cite

@article{arxiv.2001.05989,
  title  = {Conformal e-prediction},
  author = {Vladimir Vovk},
  journal= {arXiv preprint arXiv:2001.05989},
  year   = {2025}
}

Comments

Minor improvements since the previous version; 30 pages and 1 figure