Title :
Kernelized Locality-Sensitive Hashing
Author :
Kulis, Brian ; Grauman, Kristen
Author_Institution :
Comput. Sci. & Eng. Dept., Ohio State Univ., Columbus, OH, USA
fDate :
6/1/2012 12:00:00 AM
Abstract :
Fast retrieval methods are critical for many large-scale and data-driven vision applications. Recent work has explored ways to embed high-dimensional features or complex distance functions into a low-dimensional Hamming space where items can be efficiently searched. However, existing methods do not apply for high-dimensional kernelized data when the underlying feature embedding for the kernel is unknown. We show how to generalize locality-sensitive hashing to accommodate arbitrary kernel functions, making it possible to preserve the algorithm´s sublinear time similarity search guarantees for a wide class of useful similarity functions. Since a number of successful image-based kernels have unknown or incomputable embeddings, this is especially valuable for image retrieval tasks. We validate our technique on several data sets, and show that it enables accurate and fast performance for several vision problems, including example-based object classification, local feature matching, and content-based retrieval.
Keywords :
content-based retrieval; file organisation; image classification; image matching; image retrieval; arbitrary kernel function; content-based retrieval; example-based object classification; fast retrieval method; high-dimensional kernelized data; image retrieval task; image search; image-based kernel; kernelized locality-sensitive hashing; local feature matching; similarity function; sublinear time similarity search guarantees; vision problem; Approximation algorithms; Covariance matrix; Databases; Kernel; Measurement; Object recognition; Vectors; Kernel methods; Similarity search; central limit theorem; image search.; locality-sensitive hashing;
Journal_Title :
Pattern Analysis and Machine Intelligence, IEEE Transactions on
DOI :
10.1109/TPAMI.2011.219