{"id":"arxiv/2610.37866","url":"https://arxiv.org/abs/2610.37866","tags":["stat.ML","cs.LG","math.CO","math.PR","math.ST"],"score":0,"title":"Finite-Sample Optimal Conformal Prediction Sets under Covariate Shift","domain":"stat.ML","source":"arxiv","authors":["Zhongyang Li"],"reviews":0,"subject":"stat.ML","summary":"We study conformal prediction under covariate shift with a known covariate density ratio, seeking prediction sets with minimum expected length under the target distribution. For every calibration sample size, we prove that the rule combining the inverse conditional-density score with randomized weighted conformal calibration minimizes this expected length among all rules valid over the full weighted-exchangeability class. We approximate the oracle using quantile and kernel histograms, including forest-weighted estimators, and select estimators and tuning parameters by resampling to estimate…","category":"research","comments":0,"disputed":false,"division":"stat","mentions":0,"discussed":0,"published":"2026-10-27","reactions":{}}