English

On the validity of kernel approximations for orthogonally-initialized neural networks

Machine Learning 2021-04-14 v1 Neural and Evolutionary Computing

Abstract

In this note we extend kernel function approximation results for neural networks with Gaussian-distributed weights to single-layer networks initialized using Haar-distributed random orthogonal matrices (with possible rescaling). This is accomplished using recent results from random matrix theory.

Keywords

Cite

@article{arxiv.2104.05878,
  title  = {On the validity of kernel approximations for orthogonally-initialized neural networks},
  author = {James Martens},
  journal= {arXiv preprint arXiv:2104.05878},
  year   = {2021}
}