Kernelized locality-sensitive hashing for scalable image search

By Brian Kulis, Kristen Grauman. 2009

965 citations recorded by the source corpus.

Subfields: Computer Vision and Pattern Recognition, Information Retrieval & Search

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computer Vision and Pattern Recognition

In Information Retrieval & Search

Citations in the corpus

Aggregating Local Image Descriptors into Compact Codes (2012, methodology): “One of the most popular techniques is Euclidean Locality-Sensitive Hashing [36], which has been extended in [17] to arbitrary metrics.”
Product Quantization for Nearest Neighbor Search (2011, background): “Most of the effort has been devoted to the Euclidean distance, though recent generalizations have been proposed for other metrics [10].”
Learning Binary Codes for Maximum Inner Product Search (2015, background): “The LSH family has been continuously developed to accommodate a variety of distance and similarity measures, such as Euclidean distance, p-norm distance [3], Mahalanobis distance [15], kernel similarity [14, 28].”
Supervised Discrete Hashing (2015, methodology): “In addition to traditional Euclidean distance, LSH has been generalized to accommodate other distance and similarity measures such as p-norm distance [4], Mahalanobis metric [15], and kernel similarity [14, 27].”
Feature Learning Based Deep Supervised Hashing with Pairwise Labels (2015, methodology): “Existing hashing methods can be divided into dataindependent methods and data-dependent methods [9, 18, 3].”
Simultaneous feature learning and hash coding with deep neural networks (2015, methodology): “Notable examples in this category include Kernelized Locality-Sensitive Hashing [9], Semantic Hashing [19], graph-based hashing methods [26, 13], and Iterative Quantization [4].”
Supervised Hashing for Image Retrieval via Image Representation Learning (2014, methodology): “However, LSH methods usually require long hash codes to achieve satisfactory search accuracies, resulting in large space for storage and low recall in search (Liu et al. 2012).”
Polysemous Codes (2016, methodology): “The ﬁrst class of methods proposes to map the original vectors to the Hamming hypercube [38,52,34,50].”
Spherical hashing (2012, methodology): “However, previous techniques using kernel spaces [17, 10, 15] define their kernels with randomly chosen multiple landmarks from the input data points.”
Query-driven iterated neighborhood graph search for large scale indexing (2012, background): “Other examples include ﬁnding the best matches for local image features in large data sets [37] and large scale similar image search [22].”
Hashing for Distributed Data (2015, background): “Early works, such as locality sensitive hashing (LSH) (Indyk & Motwani, 1998; Charikar, 2002) and its variants (Kulis & Grauman, 2009; Datar et al., 2004), construct hash functions based on random projections without considering the input data.”
Fast Exact Max-Kernel Search (2012, background): “nctions which admit a locality sensitive hash1The last two functions provide matching scores for pairs of sequences and can easily be shown to be Mercer kernels. ing function [12]2. Kulis and Grauman [13] apply LSH to solve MKS approximately for normalized kernels without any explicit representation”
Deep hashing for compact binary codes learning (2015, methodology): “Representative methods in this category are locality sensitive hashing (LSH) [1] and its kernelized or discriminative extensions [11] [16] [23].”
Learning to Hash for Indexing Big Data—A Survey (2015, methodology): “By converting rank lists to a triplet tensor matrix, listwise hashing is designed to preserve the ranking in the Hamming space [76].”
Hashing for Similarity Search: A Survey (2014, positive): “Kernel LSH [64], [65] aims to build LSH functions with the angle defined in the kernel space, θ(xi,xj) = arccos φ(xi) φ(xj) ‖φ(xi)‖2‖φ(xj)‖2 .”
Iterative Quantization: A Procrustean Approach to Learning Binary Codes for Large-Scale Image Retrieval (2013, neutral): “R ECENTLY, the vision community has devoted a lot of attention to the problem of learning similaritypreserving binary codes for representing large-scale image collections [4], [5], [13], [15], [20], [22], [23], [29], [31], [36], [39], [43], [45], [49], [50], [51].”
Supervised Text-based Geolocation Using Language Models on an Adaptive Grid (2012, methodology): “It would also require efﬁcient similarity comparisons, using techniques such as locality-sensitive hashing (Kulis and Grauman, 2009).”