A sequential quadratic Hamiltonian algorithm for training explicit RK neural networks

DOI10.1016/j.cam.2021.113943zbMath1480.49004WikidataQ114201947 ScholiaQ114201947MaRDI QIDQ2068636

Yanyan Li

Publication date: 20 January 2022

Published in: Journal of Computational and Applied Mathematics (Search for Journal in Brave)

Full work available at URL: https://doi.org/10.1016/j.cam.2021.113943

zbMATH Keywords

numerical optimization; method of successive approximations; residual neural networks; discrete Pontryagin maximum principle; Runge-Kutta neural networks; sequential quadratic Hamiltonian algorithm

Mathematics Subject Classification ID

65K10: Numerical optimization and variational techniques

68T05: Learning and adaptive systems in artificial intelligence

49M05: Numerical methods based on necessary conditions

92B20: Neural networks for/in biological studies, artificial life and related topics

49J15: Existence theories for optimal control problems involving ordinary differential equations

49K15: Optimality conditions for problems involving ordinary differential equations

Uses Software

• LBFGS-B
• torchdiffeq