Welfare-Optimal Classification with Accuracy Auctions

2026-06-01

AI Summary

Standard classification models aim to maximize predictive accuracy but overlook the heterogeneous benefits users derive from correct predictions, thereby failing to achieve socially optimal outcomes. This work introduces mechanism design into classification learning and proposes an incentive-compatible accuracy auction mechanism that jointly optimizes social welfare and predictive performance while eliciting truthful reporting of users’ private valuations. The approach features an auction-based learning algorithm that efficiently computes allocations and payments, and theoretical analysis establishes that the number of paying users is independent of sample size. Experiments on both real-world and synthetic datasets validate the algorithm’s effectiveness and uncover a nontrivial trade-off between social welfare and prediction accuracy.

Abstract

Prediction algorithms are increasingly used to inform decisions about humans, but maximizing accuracy$\rule[0.25em]{1em}{0.4pt}$the standard learning objective$\rule[0.25em]{1em}{0.4pt}$does not necessarily maximize user benefits. Instead, we propose optimizing social welfare, defined as the average gain users receive from correct predictions. Welfare enables to express, and therefore account for, heterogeneity in how much users benefit from accuracy. But since these valuations are private and users can gain from overreporting them, learning must simultaneously elicit truthful values and optimize welfare with respect to them. To this end, we propose a novel learning algorithm that incorporates a truthful auction. We show how to compute allocations and prices efficiently, and bound the number of paying users$\rule[0.25em]{1em}{0.4pt}$ which surprisingly is independent of the sample size. We conclude with experiments on real and synthetic data that demonstrate our algorithm and explore the connections between welfare and accuracy.

Problem

• social welfare
• prediction algorithms
• accuracy
• truthful elicitation
• heterogeneous benefits

Innovation

• welfare optimization
• accuracy auctions
• truthful mechanism
• prediction algorithms
• social welfare