{"id":"arxiv/2610.35821","url":"https://arxiv.org/abs/2610.35821","tags":["stat.ME","math.ST","stat.ML"],"score":0,"title":"Recertifying decisions via e-closure: uniform improvements of conformal risk control and randomized e-BH","domain":"stat.ME","source":"arxiv","authors":["Tian Bai","Aaditya Ramdas"],"reviews":0,"subject":"stat.ME","summary":"Statistical procedures often control expected loss at a prescribed level while leaving part of their risk budget unused. Building on the e-value representation of risk-controlled decisions and the e-closure principle for multiple testing, we formulate recertification, a general paradigm for improving procedures by exploiting computable slack in their validity certificates. A valid procedure's losses normalized by their risk budgets form a state-indexed e-value collection; conversely, decisions whose normalized losses are bounded by this collection inherit its risk guarantees. Recertification…","category":"research","comments":0,"disputed":false,"division":"stat","mentions":0,"discussed":0,"published":"2026-10-26","reactions":{}}