Sparsity and Fairness in logistic regression via Penalized Mean Field Variational Inference

E. Carrizosa Priego, R. Jiménez Llamas, P. Ramírez Cobo

We propose a novel approach to Bayesian logistic regression that produces sparse and fair solutions. To do so we modify the Mean Field Variational Inference approach by adding a penalization term dependent on the unfairness of the posterior predictions and by an appropriate prior selection. This leads to a set of CAVI equations whose solutions ensure sparsity and fairness on the predictions in a private manner. Numerical results show how this new method results in a trade-off between sparsity, fairness and accuracy, which eases the decision making process for a potential user.

Keywords: Bayesian logistic regression, Sparsity, fairness, variational inference

Young Researchers in Bayesian Statistics
June 10, 2025  7:00 PM
MR 3

M. Santos Pascual, D. Ríos, P. Ángulo

D. Mlynarczyk, G. Calvo, F. Palmí Perales, C. Armero, V. Gómez-Rubio, Ú. Martínez-Iranzo, A. de la Torre-García

C. Mulet, G. García-Donato, M. E. Castellanos