An exact four-regime frontier for robust relative fairness with selective labels

This technical note determines the minimum additive tolerance needed for a same-world, benchmark-relative false-positive-rate contraction guarantee under selective labels while requiring a fixed positive accuracy gain for each group in every compatible completion. For a four-cell population model with a=7/10, c=1/10, label-revelation probability s=99/100 in the first three cells, and a wholly unlabeled fourth cell, it gives the exact four-piece frontier for every contraction coefficient rho in [0,1] over all group-blind randomized rules q in [0,1]^4, at guaranteed gain gamma=1/25. The attaining rule is unique throughout. The optimal second-cell probability first increases and then decreases together with the fourth-cell action probability as contraction strengthens. Three actual compatible world pairs give sharp lower bounds, and full-box sign conditions establish attainment. The note also proves a two-way normalized-oscillation representation. The statements are conditional population results for the specified rectangular completion family.

Authors

Institutions

Publication Details

Journal
Zenodo (CERN European Organization for Nuclear Research)
Published
2026-10-27
DOI
https://doi.org/10.5281/zenodo.23680052
Primary Topic
Ethics and Social Impacts of AI
Type
article
Field-Weighted Citation Impact
0.00
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
article

An exact four-regime frontier for robust relative fairness with selective labels

Zihuan Xu
Zenodo (CERN European Organization for Nuclear Research)
Ethics and Social Impacts of AI
article

An exact four-regime frontier for robust relative fairness with selective labels

Zihuan Xu
article en

Abstract

This technical note determines the minimum additive tolerance needed for a same-world, benchmark-relative false-positive-rate contraction guarantee under selective labels while requiring a fixed positive accuracy gain for each group in every compatible completion. For a four-cell population model with a=7/10, c=1/10, label-revelation probability s=99/100 in the first three cells, and a wholly unlabeled fourth cell, it gives the exact four-piece frontier for every contraction coefficient rho in [0,1] over all group-blind randomized rules q in [0,1]^4, at guaranteed gain gamma=1/25. The attaining rule is unique throughout. The optimal second-cell probability first increases and then decreases together with the fourth-cell action probability as contraction strengthens. Three actual compatible world pairs give sharp lower bounds, and full-box sign conditions establish attainment. The note also proves a two-way normalized-oscillation representation. The statements are conditional population results for the specified rectangular completion family.

Zenodo (CERN European Organization for Nuclear Research)
Capital Normal University (CN)
Openalex Percentile: Top 8%
Ethics and Social Impacts of AI
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon