English

A matrix concentration inequality for products

Probability 2020-08-20 v2 Machine Learning

Abstract

We present a non-asymptotic concentration inequality for the random matrix product \begin{equation}\label{eq:Zn} Z_n = \left(I_d-\alpha X_n\right)\left(I_d-\alpha X_{n-1}\right)\cdots \left(I_d-\alpha X_1\right), \end{equation} where {Xk}k=1+\left\{X_k \right\}_{k=1}^{+\infty} is a sequence of bounded independent random positive semidefinite matrices with common expectation E[Xk]=Σ\mathbb{E}\left[X_k\right]=\Sigma. Under these assumptions, we show that, for small enough positive α\alpha, ZnZ_n satisfies the concentration inequality \begin{equation}\label{eq:CTbound} \mathbb{P}\left(\left\Vert Z_n-\mathbb{E}\left[Z_n\right]\right\Vert \geq t\right) \leq 2d^2\cdot\exp\left(\frac{-t^2}{\alpha \sigma^2} \right) \quad \text{for all } t\geq 0, \end{equation} where σ2\sigma^2 denotes a variance parameter.

Keywords

Cite

@article{arxiv.2008.05104,
  title  = {A matrix concentration inequality for products},
  author = {Sina Baghal},
  journal= {arXiv preprint arXiv:2008.05104},
  year   = {2020}
}
R2 v1 2026-06-23T17:47:49.926Z