English

Concentration of measure for non-linear random matrices with applications to neural networks and non-commutative polynomials

Probability 2025-07-15 v2 Machine Learning

Abstract

We prove concentration inequalities for several models of non-linear random matrices. As corollaries we obtain estimates for linear spectral statistics of the conjugate kernel of neural networks and non-commutative polynomials in (possibly dependent) random matrices.

Keywords

Cite

@article{arxiv.2507.07625,
  title  = {Concentration of measure for non-linear random matrices with applications to neural networks and non-commutative polynomials},
  author = {Radosław Adamczak},
  journal= {arXiv preprint arXiv:2507.07625},
  year   = {2025}
}

Comments

Some typos fixed (and some new probably introduced), small editorial changes