Fast metric embedding into the Hamming cube

TL;DR

This paper introduces a fast, data-oblivious method for embedding high-dimensional Euclidean data into a low-dimensional Hamming cube with near-isometric properties, using structured random matrices and binarization.

Contribution

It presents a novel, efficient embedding technique utilizing double circulant matrices that mimic Gaussian behavior, improving upon previous methods in speed and theoretical guarantees.

Findings

Achieves near-isometric embedding with high probability

Uses structured random matrices for fast computation

Provides optimal bounds on bits needed for encoding

Abstract

We consider the problem of embedding a subset of $\mathbb{R}^n$ into a low-dimensional Hamming cube in an almost isometric way. We construct a simple, data-oblivious, and computationally efficient map that achieves this task with high probability: we first apply a specific structured random matrix, which we call the double circulant matrix; using that matrix requires linear storage and matrix-vector multiplication can be performed in near-linear time. We then binarize each vector by comparing each of its entries to a random threshold, selected uniformly at random from a well-chosen interval. We estimate the number of bits required for this encoding scheme in terms of two natural geometric complexity parameters of the set - its Euclidean covering numbers and its localized Gaussian complexity. The estimate we derive turns out to be the best that one can hope for - up to logarithmic terms. The key to the proof is a phenomenon of independent interest: we show that the double circulant matrix mimics the behavior of a Gaussian matrix in two important ways. First, it maps an arbitrary set in $\mathbb{R}^n$ into a set of well-spread vectors. Second, it yields a fast near-isometric embedding of any finite subset of $\ell_2^n$ into $\ell_1^m$. This embedding achieves the same dimension reduction as a Gaussian matrix in near-linear time, under an optimal condition - up to logarithmic factors - on the number of points to be embedded. This improves a well-known construction due to Ailon and Chazelle.