Learning Space Partitions for Nearest Neighbor Search

TL;DR

This paper introduces Neural Locality-Sensitive Hashing (Neural LSH), a novel framework for space partitioning in nearest neighbor search that outperforms traditional methods on standard benchmarks.

Contribution

The paper develops a new framework for space partitioning in NNS by combining graph partitioning and neural networks, leading to improved partitioning methods.

Findings

Neural LSH outperforms quantization and tree-based methods.

Neural LSH surpasses classic data-oblivious LSH.

Experimental results show consistent improvements on benchmarks.

Abstract

Space partitions of $\mathbb{R}^d$ underlie a vast and important class of fast nearest neighbor search (NNS) algorithms. Inspired by recent theoretical work on NNS for general metric spaces [Andoni, Naor, Nikolov, Razenshteyn, Waingarten STOC 2018, FOCS 2018], we develop a new framework for building space partitions reducing the problem to balanced graph partitioning followed by supervised classification. We instantiate this general approach with the KaHIP graph partitioner [Sanders, Schulz SEA 2013] and neural networks, respectively, to obtain a new partitioning procedure called Neural Locality-Sensitive Hashing (Neural LSH). On several standard benchmarks for NNS, our experiments show that the partitions obtained by Neural LSH consistently outperform partitions found by quantization-based and tree-based methods as well as classic, data-oblivious LSH.