Exploiting Modern Hardware for High-Dimensional Nearest Neighbor Search

TL;DR

This paper explores how to leverage modern CPU features like SIMD and cache hierarchies to enhance the efficiency of product quantization-based high-dimensional nearest neighbor search in large-scale multimedia and machine learning applications.

Contribution

It introduces techniques that exploit modern hardware capabilities to significantly improve response times of product quantization methods for high-dimensional nearest neighbor search.

Findings

Enhanced response times using SIMD instructions.

Efficient utilization of CPU cache hierarchy.

Scalability to large multimedia databases.

Abstract

Many multimedia information retrieval or machine learning problems require efficient high-dimensional nearest neighbor search techniques. For instance, multimedia objects (images, music or videos) can be represented by high-dimensional feature vectors. Finding two similar multimedia objects then comes down to finding two objects that have similar feature vectors. In the current context of mass use of social networks, large scale multimedia databases or large scale machine learning applications are more and more common, calling for efficient nearest neighbor search approaches. This thesis builds on product quantization, an efficient nearest neighbor search technique that compresses high-dimensional vectors into short codes. This makes it possible to store very large databases entirely in RAM, enabling low response times. We propose several contributions that exploit the capabilities of modern CPUs, especially SIMD and the cache hierarchy, to further decrease response times offered by product quantization.