AI Paper Atlas

Research map / Learning theory and algorithms / Fairness

Beyond Procedure: Substantive Fairness in Conformal Prediction

Pengqi Liu, Zijun Yu, Mouloud Belbahri, Arthur Charpentier, Masoud Asgharian, Jesse C. Cresswell

ICML 2026 · Learning theory and algorithms › Fairness · approach: Fairness auditing and regularization · 0 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
use Mondrian (group-conditional) CP instead of Marginal CPCP with LLM-in-the-loop · FACETmaxROR (%): 9.0 -> 38Hurt
use Mondrian (group-conditional) CP instead of Marginal CPCP with LLM-in-the-loop · RAVDESSmaxROR (%): 11 -> 79Hurt
use Mondrian (group-conditional) CP instead of Marginal CPCP with LLM-in-the-loop · BiosBiasmaxROR (%): 6.9 -> 8.1Hurt
use Mondrian (group-conditional) CP instead of Marginal CPCP with human-in-the-loop · FACETmaxROR (%): 26 -> 51Hurt
use Mondrian CP instead of Marginal CPCP with human-in-the-loop · RAVDESSmaxROR (%): 1.0 -> 28Hurt
use Mondrian CP instead of Marginal CPbase classifier XGBoost · ACSIncomeaverage set size: 5.35 -> 7.16Hurt
use Backward CP instead of Marginal CPbase classifier CLIP ViT-L/14 · FACETaverage set size: 2.62 -> 3.50Hurt
equalize coverage between groups (reduce coverage gap)CP methods · all four tasksmaxROR (%)Hurt
equalize set size between groups (reduce set size gap)CP methods · all four tasksmaxROR (%)Improved
use Label-Clustered CP instead of Marginal CPbase classifier XGBoost · ACSIncomeaverage set size: 5.35 -> 5.33No clear effect
use Label-Clustered CP with K=2 clusters instead of K=1 (Marginal) (ablation)Label-Clustered CP · BiosBias and RAVDESSset size gapImproved
increase number of label clusters K beyond the minimum (ablation)Label-Clustered CP · RAVDESSset size gapHurt

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy