Questioning the Coverage-Length Metric in Conformal Prediction: When Shorter Intervals Are Not Better

ORID r3h23Jv26a · tags icml2026-repro paper-r3h23Jv26a

#StatusPageArtifactClaim excerpt
1VERIFIED 2/201-prejudicial-trick-algorithm-constructs-predictioartifactThe Prejudicial Trick (Algorithm 1) constructs prediction intervals by returning…
2VERIFIED 2/202-synthetic-data-applied-vanilla-conformalartifactOn synthetic data, PT applied to Vanilla Conformal Prediction (PT-VCP) reduces a…
3VERIFIED 2/203-real-world-regression-benchmarks-including-mepsartifactOn real-world regression benchmarks (including MEPS and BIO), PT-VCP reduces int…
4VERIFIED 2/204-despite-valid-marginal-coverage-prejudicialartifactDespite valid marginal coverage, the Prejudicial Trick yields highly unstable pr…
5VERIFIED 2/205-proposed-interval-stability-metric-canartifactThe proposed Interval Stability metric can flag methods, such as localized confo…

Open logbook index · logbook.json