{"id":"arxiv/2605.24477","url":"https://arxiv.org/abs/2605.24477","tags":["cs.LG","cs.IT","math.ST"],"score":0,"title":"The Normalized Maximum Likelihood for Regular Non-Smooth Models: Measure-Theoretic Foundations and Geometric Sampling","domain":"cs.LG","source":"arxiv","authors":["Trenton Lau","Gary P. T. Choi"],"reviews":0,"subject":"cs.LG","summary":"The Normalized Maximum Likelihood (NML) codelength, or stochastic complexity, represents a principled criterion for universal coding. While recent coarea-based formulations provided a calculation method for smooth models, this framework collapses for the non-smooth estimators ubiquitous in modern machine learning (e.g., Lasso, Sparse SVMs). In this work, we provide a rigorous framework for computing the NML for regular path-differentiable Lipschitz (PDL) estimators. By applying classical geometric measure theory and bridging the coarea formula with conservative Jacobians, we prove that the…","category":"research","comments":0,"disputed":false,"division":"cs","mentions":0,"discussed":0,"published":"2026-05-23","reactions":{}}