{"ID":2828437,"CreatedAt":"2026-06-01T04:54:23.091178241Z","UpdatedAt":"2026-06-01T04:54:23.091178241Z","DeletedAt":null,"paper_url":"https://arxiv.org/abs/2512.14246","arxiv_id":"2512.14246","title":"Randomized multi-class classification under system constraints: a unified approach via post-processing","abstract":"We study the problem of multi-class classification under system-level constraints expressible as linear functionals over randomized classifiers. We propose a post-processing approach that adjusts a given base classifier to satisfy general constraints without retraining. Our method formulates the problem as a linearly constrained stochastic program over randomized classifiers, and leverages entropic regularization and dual optimization techniques to construct a feasible solution. We provide finite-sample guarantees for the risk and constraint satisfaction for the final output of our algorithm under minimal assumptions. The framework accommodates a broad class of constraints, including fairness, abstention, and churn requirements.","short_abstract":"We study the problem of multi-class classification under system-level constraints expressible as linear functionals over randomized classifiers. We propose a post-processing approach that adjusts a given base classifier to satisfy general constraints without retraining. Our method formulates the problem as a linearly c...","url_abs":"https://arxiv.org/abs/2512.14246","url_pdf":"https://arxiv.org/pdf/2512.14246v1","authors":"[\"Evgenii Chzhen\",\"Mohamed Hebiri\",\"Gayane Taturyan\"]","published":"2025-12-16T09:53:22Z","proceeding":"math.OC","tasks":"[\"math.OC\",\"stat.ML\"]","methods":"[]","has_code":false}