We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:

Download:

Current browse context:

math.OC

Change to browse by:

References & Citations

Bookmark

(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo ScienceWISE logo

Mathematics > Optimization and Control

Title: A Measure Theoretical Approach to the Mean-field Maximum Principle for Training NeurODEs

Abstract: In this paper we consider a measure-theoretical formulation of the training of NeurODEs in the form of a mean-field optimal control with $L^2$-regularization of the control. We derive first order optimality conditions for the NeurODE training problem in the form of a mean-field maximum principle, and show that it admits a unique control solution, which is Lipschitz continuous in time. As a consequence of this uniqueness property, the mean-field maximum principle also provides a strong quantitative generalization error for finite sample approximations. Our derivation of the mean-field maximum principle is much simpler than the ones currently available in the literature for mean-field optimal control problems, and is based on a generalized Lagrange multiplier theorem on convex sets of spaces of measures. The latter is also new, and can be considered as a result of independent interest.
Subjects: Optimization and Control (math.OC)
Cite as: arXiv:2107.08707 [math.OC]
  (or arXiv:2107.08707v2 [math.OC] for this version)

Submission history

From: Cristina Cipriani [view email]
[v1] Mon, 19 Jul 2021 09:31:33 GMT (2534kb,D)
[v2] Fri, 8 Apr 2022 15:22:04 GMT (11657kb,D)

Link back to: arXiv, form interface, contact.