{"id":"arxiv/2603.20025","url":"https://arxiv.org/abs/2603.20025","tags":["stat.ML","cs.LG","math.ST"],"score":0,"title":"Graph-Informed Adversarial Modeling: Infimal Subadditivity of Interpolative Divergences","domain":"stat.ML","source":"arxiv","authors":["Panagiota Birmpa","Eric Joseph Hall"],"reviews":0,"subject":"stat.ML","summary":"We study adversarial learning when the target distribution factorizes according to a known Bayesian network. For interpolative divergences, including $(f,Γ)$-divergences, we prove a new infimal subadditivity principle showing that, under suitable conditions, a global variational discrepancy is controlled by an average of family-level discrepancies aligned with the graph. In an additive regime, the surrogate is exact. This closes a theoretical gap in the literature; existing subadditivity results justify graph-informed adversarial learning for classical discrepancies, but not for…","category":"research","comments":0,"disputed":false,"division":"stat","mentions":0,"discussed":0,"published":"2026-03-20","reactions":{}}