{"id":"arxiv/2609.17629","url":"https://arxiv.org/abs/2609.17629","tags":["math.OC","math.ST","stat.ML"],"score":0,"title":"Fenchel-Young Duality Gaps: Certified Early Stopping for Regularized Inverse Problems","domain":"math.OC","source":"arxiv","authors":["Pierre-Cyril Aubin-Frankowski","Yohann de Castro"],"reviews":0,"subject":"math.OC","summary":"We study computable error bounds and certified early stopping for regularized inverse problems, where a data-fidelity term is traded against a regularizer. The analysis relies on an exact duality-gap identity that splits the total gap of $F(Φμ)+λR(μ)$ into a data-fidelity Fenchel--Young loss and a regularizer Fenchel--Young loss, $ Δ(μ,h)=L_F(Φμ\\parallel h)+λL_R(μ\\parallelη),\\qquad η=-Φ^\\star h/λ, $ valid for any primal point $μ$ and any dual point $h$. The data-fidelity term $F$ is strictly convex, so wherever $F^\\star$ is differentiable the loss $L_F(Φμ\\parallel h)$ is the Bregman…","category":"research","comments":0,"disputed":false,"division":"math","mentions":0,"discussed":0,"published":"2026-09-15","reactions":{}}