Id: arxiv/2603.20135 Url: https://arxiv.org/abs/2603.20135 Tags: math.ST, cs.IT Score: 0 Title: Classifier-Based Nonparametric Sequential Hypothesis Testing Domain: math.ST Source: arxiv Authors: Chia-Yu Hsu, Shubhanshu Shekhar Reviews: 0 Subject: math.ST Category: research Comments: 0 Disputed: false Division: math Mentions: 0 Discussed: 0 Published: 2026-03-20 Page: https://delphi.science/r/arxiv/2603.20135 We consider the problem of constructing sequential power-one tests where the null and alternative classes are specified indirectly through historical or offline data. More specifically, given an offline dataset consisting of observations from $L+1$ distributions $\{P_0, P_1, \ldots, P_L\}$, and a new unlabeled data stream $\{X_t: t \geq 1\} \overset{i.i.d}{\sim} P_θ$, the goal is to decide between the null $H_0: θ= 0$, against the alternative $H_1: θ\in [L]:=\{1,\ldots,L\}$. Our main methodological contribution is a general approach for designing a level-$α$ power-one test for this problem…