Research map / Learning theory and algorithms / Fairness

Selective Omniprediction and Fair Abstention

NeurIPS 2025 · Learning theory and algorithms › Fairness · approach: Fairness-constrained classification · 1 citations (OpenAlex)

Read the paperCheck your idea against 108,967 papers

AI-extracted summary. Written by AI from the paper's abstract only. Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Proposes learning algorithms for selective classifiers that abstain at a fixed cost, building on multigroup fairness and omniprediction to produce a single classifier that works across a pre-specified class of losses.

Problem

Building selective classifiers that may abstain on part of the domain at a fixed cost, while remaining competitive across a pre-specified class of loss functions and respecting multigroup fairness.

Approach

Extends the multigroup fairness and omniprediction framework to selective prediction with abstention, giving an algorithm that outputs a single classifier with abstention that is simultaneously competitive for a given class of loss functions.

Experimental setup

Builds on: multigroup fairness omniprediction

Closest papers by meaning

• Multigroup Fairness and Omniprediction: Separations and Equivalences (NeurIPS 2026 · similarity 0.89)
• Individual Fairness In Strategic Classification (NeurIPS 2025 · similarity 0.87)
• Conformal Classification with Equalized Coverage for Adaptively Selected Groups (NeurIPS 2024 · similarity 0.86)
• Fairness-aware Bayes Optimal Functional Classification (NeurIPS 2025 · similarity 0.86)
• Online Multi-Class Selection with Group Fairness Guarantee (NeurIPS 2025 · similarity 0.86)
• Accuracy vs. Accuracy: Computational Tradeoffs Between Classification Rates and Utility (NeurIPS 2026 · similarity 0.86)