Criteria of efficiency for conformal prediction

Abstract

We study optimal conformity measures for various criteria of efficiency of classification in an idealised setting. This leads to an important class of criteria of efficiency that we call probabilistic; it turns out that the most standard criteria of efficiency used in literature on conformal prediction are not probabilistic unless the problem of classification is binary. We consider both unconditional and label-conditional conformal prediction.

31 pages

Cited by in corpus (4)

• Conformal Prediction: a Unified Review of Theory and New Challenges
• Well-calibrated Confidence Measures for Multi-label Text Classification with a Large Number of Labels
• Android Malware Detection with Unbiased Confidence Guarantees
• Concepts and Applications of Conformal Prediction in Computational Drug Discovery