{"id":"arxiv/2606.00469","url":"https://arxiv.org/abs/2606.00469","tags":["math.OC","stat.ML"],"score":0,"title":"Constructive interpolation and generalization rates for neural ODEs: a control perspective","domain":"math.OC","source":"arxiv","authors":["Antonio Álvarez-López","Lorenzo Liverani","Enrique Zuazua"],"reviews":0,"subject":"math.OC","summary":"We study supervised regression with neural ODEs (NODEs) from a control-theoretic perspective to derive explicit population-risk bounds. We focus on a widely used class of non-autonomous models with constant parameters and explicit time dependence, which we call semi-autonomous NODEs (SA-NODEs). We constructively prove that SA-NODEs are capable of \\emph{exact} interpolation of admissible finite datasets, and even satisfy a stronger property that we call \\emph{simultaneous cell controllability} (SCC): their flows can map prescribed disjoint cells into arbitrarily small target balls. This…","category":"research","comments":0,"disputed":false,"division":"math","mentions":0,"discussed":0,"published":"2026-05-30","reactions":{}}