Title :
Optimized Product Quantization for Approximate Nearest Neighbor Search
Author :
Tiezheng Ge ; Kaiming He ; Qifa Ke ; Jian Sun
Author_Institution :
Univ. of Sci. & Technol. of China, Hefei, China
Abstract :
Product quantization is an effective vector quantization approach to compactly encode high-dimensional vectors for fast approximate nearest neighbor (ANN) search. The essence of product quantization is to decompose the original high-dimensional space into the Cartesian product of a finite number of low-dimensional subspaces that are then quantized separately. Optimal space decomposition is important for the performance of ANN search, but still remains unaddressed. In this paper, we optimize product quantization by minimizing quantization distortions w.r.t. the space decomposition and the quantization codebooks. We present two novel methods for optimization: a non-parametric method that alternatively solves two smaller sub-problems, and a parametric method that is guaranteed to achieve the optimal solution if the input data follows some Gaussian distribution. We show by experiments that our optimized approach substantially improves the accuracy of product quantization for ANN search.
Keywords :
Gaussian distribution; computer vision; optimisation; search problems; vector quantisation; ANN search; Cartesian product; Gaussian distribution; approximate nearest neighbor search; computer vision; high-dimensional vectors; low-dimensional subspaces; nonparametric method; optimal space decomposition; optimization; optimized product quantization; parametric method; quantization codebooks; quantization distortions; space decomposition; vector quantization approach; Artificial neural networks; Eigenvalues and eigenfunctions; Encoding; Linear programming; Optimization; Quantization (signal); Vectors; nearest neighbor search; product quantization;
Conference_Titel :
Computer Vision and Pattern Recognition (CVPR), 2013 IEEE Conference on
Conference_Location :
Portland, OR
DOI :
10.1109/CVPR.2013.379