Fourier-domain Variational Formulation and Its Well-posedness for Supervised Learning
From MaRDI portal
Publication:6355404
arXiv2012.03238MaRDI QIDQ6355404FDOQ6355404
Authors: Tao Luo, Zheng Ma, Zhiwei Wang, Zhi-Qin John Xu, Yaoyu Zhang 
Publication date: 6 December 2020
Abstract: A supervised learning problem is to find a function in a hypothesis function space given values on isolated data points. Inspired by the frequency principle in neural networks, we propose a Fourier-domain variational formulation for supervised learning problem. This formulation circumvents the difficulty of imposing the constraints of given values on isolated data points in continuum modelling. Under a necessary and sufficient condition within our unified framework, we establish the well-posedness of the Fourier-domain variational problem, by showing a critical exponent depending on the data dimension. In practice, a neural network can be a convenient way to implement our formulation, which automatically satisfies the well-posedness condition.
Numerical optimization and variational techniques (65K10) Numerical methods for ill-posed problems for boundary value problems involving PDEs (65N20)
This page was built for publication: Fourier-domain Variational Formulation and Its Well-posedness for Supervised Learning
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q6355404)
