Free denoising via overlap measures and c-freeness techniques

TL;DR

This paper introduces a novel free denoising method using overlap measures and c-freeness techniques to estimate signals from noisy free random variables, with applications to both additive and multiplicative noise models.

Contribution

It develops a new framework for free denoising based on overlap measures and c-freeness, connecting free probability with matrix denoising methods.

Findings

Overlap measure $ ext{μ}^{( ext{ov})}_{a,a+b}$ is absolutely continuous w.r.t. product measure.

Radon-Nikodym derivative provides direct access to $E(a|a+b)$.

Results extend to multiplicative noise and general selfadjoint expressions.

Abstract

We study the problem of free denoising. For free selfadjoint random variables $a,b$, where we interpret $a$ as a signal and $b$ as noise, we find $E(a|a+b)$. To that end, we study a probability measure $\mu^{( \mathrm{ov} )}_{a,a+b}$ on $\mathbb{R}^2$ which we call the overlap measure. We show that $\mu^{( \mathrm{ov} )}_{a,a+b}$ is absolutely continuous with respect to the product measure $\mu_a\times \mu_{a+b}$. The Radon-Nikodym derivative gives direct access to $E(a|a+b)$. We show that analogous results hold in the case of multiplicative noise when $a,b$ are positive and the aim is to find $E(a|a^{1/2}ba^{1/2})$. In a parallel development we show that, for a general selfadjoint expression $P(a,b)$ made with $a$ and $b$, finding $E(a|P(a,b))$ is equivalent to finding the distribution of $P(a,b)$ in a certain two-state probability space $(\mathcal{A},\varphi,\chi)$, where $a,b$ are c-free with respect to $(\varphi,\chi)$ in the sense of Bo\.zejko-Leinert-Speicher. We discuss how free denoising (which is set in the framework of an abstract $W^{*}$-probability space) relates to the notion of ''matrix denoising'' previously discussed in the random matrix literature.