{"id":"arxiv/2601.07325","url":"https://arxiv.org/abs/2601.07325","tags":["stat.ML","cs.LG","math.ST"],"score":0,"title":"Robust Bayesian Inference via Variational Approximations of Generalized Rho-Posteriors","domain":"stat.ML","source":"arxiv","authors":["EL Mahdi Khribch","Pierre Alquier"],"reviews":0,"subject":"stat.ML","summary":"We introduce the $\\widetildeρ$-posterior, a modified version of the $ρ$-posterior, obtained by replacing the supremum over competitor parameters with a softmax aggregation. This modification allows a PAC-Bayesian analysis of the $\\widetildeρ$-posterior. This yields finite-sample oracle inequalities with explicit convergence rates that inherit the key robustness properties of the original framework, in particular, graceful degradation under model misspecification and data contamination. Crucially, the PAC-Bayesian oracle inequalities extend to variational approximations of the…","category":"research","comments":0,"disputed":false,"division":"stat","mentions":0,"discussed":0,"published":"2026-01-12","reactions":{}}