Efficient Sketching Algorithm for Sparse Binary Data

TL;DR

This paper introduces $insketch$, a fast and simple sketching algorithm for high-dimensional sparse binary data that preserves multiple similarity measures, enabling efficient data analysis tasks with comparable accuracy to existing methods.

Contribution

The paper proposes $insketch$, a novel sketching algorithm that efficiently reduces dimensionality of sparse binary datasets while preserving key similarity measures, with theoretical analysis and practical validation.

Findings

$insketch$ achieves similar accuracy to state-of-the-art methods.

It offers significant speedup in dimensionality reduction.

The algorithm is simple and easy to implement.

Abstract

Recent advancement of the WWW, IOT, social network, e-commerce, etc. have generated a large volume of data. These datasets are mostly represented by high dimensional and sparse datasets. Many fundamental subroutines of common data analytic tasks such as clustering, classification, ranking, nearest neighbour search, etc. scale poorly with the dimension of the dataset. In this work, we address this problem and propose a sketching (alternatively, dimensionality reduction) algorithm -- $\binsketch$ (Binary Data Sketch) -- for sparse binary datasets. $\binsketch$ preserves the binary version of the dataset after sketching and maintains estimates for multiple similarity measures such as Jaccard, Cosine, Inner-Product similarities, and Hamming distance, on the same sketch. We present a theoretical analysis of our algorithm and complement it with extensive experimentation on several real-world datasets. We compare the performance of our algorithm with the state-of-the-art algorithms on the task of mean-square-error and ranking. Our proposed algorithm offers a comparable accuracy while suggesting a significant speedup in the dimensionality reduction time, with respect to the other candidate algorithms. Our proposal is simple, easy to implement, and therefore can be adopted in practice.