Column randomization and almost-isometric embeddings

03/09/2021
by   Shahar Mendelson, et al.
0

The matrix A:ℝ^n →ℝ^m is (δ,k)-regular if for any k-sparse vector x, | Ax_2^2-x_2^2| ≤δ√(k)x_2^2. We show that if A is (δ,k)-regular for 1 ≤ k ≤ 1/δ^2, then by multiplying the columns of A by independent random signs, the resulting random ensemble A_ϵ acts on an arbitrary subset T ⊂ℝ^n (almost) as if it were gaussian, and with the optimal probability estimate: if ℓ_*(T) is the gaussian mean-width of T and d_T=sup_t ∈ Tt_2, then with probability at least 1-2exp(-c(ℓ_*(T)/d_T)^2), sup_t ∈ T| A_ϵ t_2^2-t_2^2 | ≤ C(Λ d_T δℓ_*(T)+(δℓ_*(T))^2 ), where Λ=max{1,δ^2log(nδ^2)}. This estimate is optimal for 0<δ≤ 1/√(log n).

READ FULL TEXT
POST COMMENT

Comments

There are no comments yet.

Authors

page 1

page 2

page 3

page 4

04/02/2018

Sparse Gaussian ICA

Independent component analysis (ICA) is a cornerstone of modern data ana...
02/21/2017

Column normalization of a random measurement matrix

In this note we answer a question of G. Lecué, by showing that column no...
08/25/2021

Matchings on Random Regular Hypergraphs

We prove the convergence in probability of free energy for matchings on ...
03/24/2022

On the Binary and Boolean Rank of Regular Matrices

A 0,1 matrix is said to be regular if all of its rows and columns have t...
04/08/2022

Fast metric embedding into the Hamming cube

We consider the problem of embedding a subset of ℝ^n into a low-dimensio...
02/01/2021

Optimal robust mean and location estimation via convex programs with respect to any pseudo-norms

We consider the problem of robust mean and location estimation w.r.t. an...
11/15/2017

Gaussian width bounds with applications to arithmetic progressions in random settings

Motivated by two problems on arithmetic progressions (APs)---concerning ...
This week in AI

Get the week's most popular data science and artificial intelligence research sent straight to your inbox every Saturday.