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

Download:

Current browse context:

cs.LG

Change to browse by:

References & Citations

DBLP - CS Bibliography

Bookmark

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

Computer Science > Machine Learning

Title: Born-Infeld (BI) for AI: Energy-Conserving Descent (ECD) for Optimization

Abstract: We introduce a novel framework for optimization based on energy-conserving Hamiltonian dynamics in a strongly mixing (chaotic) regime and establish its key properties analytically and numerically. The prototype is a discretization of Born-Infeld dynamics, with a squared relativistic speed limit depending on the objective function. This class of frictionless, energy-conserving optimizers proceeds unobstructed until slowing naturally near the minimal loss, which dominates the phase space volume of the system. Building from studies of chaotic systems such as dynamical billiards, we formulate a specific algorithm with good performance on machine learning and PDE-solving tasks, including generalization. It cannot stop at a high local minimum and cannot overshoot the global minimum, yielding an advantage in non-convex loss functions, and proceeds faster than GD+momentum in shallow valleys.
Comments: 9 pages + Appendix, 8 figures. Code available online
Subjects: Machine Learning (cs.LG); Cosmology and Nongalactic Astrophysics (astro-ph.CO); High Energy Physics - Theory (hep-th); Optimization and Control (math.OC); Machine Learning (stat.ML)
Cite as: arXiv:2201.11137 [cs.LG]
  (or arXiv:2201.11137v1 [cs.LG] for this version)

Submission history

From: Giuseppe Bruno De Luca [view email]
[v1] Wed, 26 Jan 2022 19:00:05 GMT (4258kb,D)

Link back to: arXiv, form interface, contact.