A Survey on Learning to Hash

TL;DR

This survey comprehensively reviews learning to hash algorithms for nearest neighbor search, categorizing methods, analyzing their performance, and discussing emerging topics in the field.

Contribution

It provides a detailed categorization of learning to hash methods, compares their performance, and discusses the relationship between different approaches and emerging trends.

Findings

Quantization algorithms outperform others in accuracy, speed, and space efficiency.

Quantization can be derived from pairwise similarity preservation.

Evaluation protocols and performance analysis are systematically presented.

Abstract

Nearest neighbor search is a problem of finding the data points from the database such that the distances from them to the query point are the smallest. Learning to hash is one of the major solutions to this problem and has been widely studied recently. In this paper, we present a comprehensive survey of the learning to hash algorithms, categorize them according to the manners of preserving the similarities into: pairwise similarity preserving, multiwise similarity preserving, implicit similarity preserving, as well as quantization, and discuss their relations. We separate quantization from pairwise similarity preserving as the objective function is very different though quantization, as we show, can be derived from preserving the pairwise similarities. In addition, we present the evaluation protocols, and the general performance analysis, and point out that the quantization algorithms perform superiorly in terms of search accuracy, search time cost, and space cost. Finally, we introduce a few emerging topics.