Active Bipartite Ranking with Smooth Posterior Distributions
James Cheshire, Stephan Clémençon · 2026
Read Active Bipartite Ranking with Smooth Posterior Distributions by James Cheshire, Stephan Clémençon (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2602-24263.txt.
Description
James Cheshire and Stephan Clémençon study active bipartite ranking with smooth posterior distributions, generalizing earlier active ranking work beyond discrete settings. The 2026 paper treats bipartite ranking, a learning problem behind many applications, in a continuous setting where conditional distributions satisfy a Hölder smoothness constraint rather than staying piecewise constant.
The authors first show that a naive approach, discretizing at a uniform level fixed in advance and applying the discrete active strategy, generally fails. Instead they propose smooth-rank, a novel algorithm that minimizes the distance between the estimated ranking rule's receiver operating characteristic curve and the optimal one under the supremum norm. For a fixed confidence level and failure probability, smooth-rank is shown to be probably approximately correct, with problem-dependent upper bounds on expected sampling time and matching lower bounds for any such algorithm. Numerical results support the theory and compare favorably with alternatives.
Catalog record
| CREATOR | James Cheshire, Stephan Clémençon |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 117 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2602-24263