EUVIMEDEUVIMED entdecken
Crossref

An Entropy Formula for the Deep Linear Network

Govind Menon, Tianmin Yu

SIAM Journal on Mathematical Analysis · 31.12.2026 · 58 (6) · 6142-6164

DOI: 10.1137/26m1847996 ↗

Kurz gefasst

Fachartikel in der Zeitschrift SIAM Journal on Mathematical Analysis, online erschienen am 05.11.2026. Verfasst von Govind Menon und 1 weiteren Autor:innen. Die Arbeit nennt 32 Literaturangaben. Gefördert von: Academia Sinica, National Science Foundation.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

We study the Riemannian geometry of the deep linear network as a foundation for a thermodynamic description of the learning process. The main tools are the use of group actions to analyze overparametrization and the use of Riemannian submersion from the space of parameters to the space of observables. The foliation of the balanced manifold in the parameter space by group orbits is used to define and compute a Boltzmann entropy. We also show that the Riemannian geometry on the space of observables defined Bah et al. [ Inform. Inference, 11 (2022), pp. 307–353 ] is obtained by Riemannian submersion of the balanced manifold. The main technical step is an explicit construction of an orthonormal basis for the tangent space of the balanced manifold using the theory of Jacobi matrices.

Förderung

  • Academia Sinica
  • National Science Foundation

Zitieren

APA

Menon, G., & Yu, T. (2026). An Entropy Formula for the Deep Linear Network. SIAM Journal on Mathematical Analysis, 58(6), 6142-6164. https://doi.org/10.1137/26m1847996

Vancouver

Menon G, Yu T. An Entropy Formula for the Deep Linear Network. SIAM Journal on Mathematical Analysis. 2026;58(6):6142-6164. doi:10.1137/26m1847996

BibTeX
@article{Menon2026,
  author = {Menon, Govind and Yu, Tianmin},
  title = {An Entropy Formula for the Deep Linear Network},
  journal = {SIAM Journal on Mathematical Analysis},
  year = {2026},
  volume = {58},
  number = {6},
  pages = {6142--6164},
  doi = {10.1137/26m1847996}
}
Kostenlos starten →