Product Quantization for Nearest Neighbor Search

TLDR

The paper proposes a product quantization method for efficient approximate nearest neighbor search. The method partitions the space into low‑dimensional subspaces, quantizes each subspace, encodes vectors with short subspace indices, and estimates Euclidean distances from these codes, with an asymmetric variant improving precision. Experiments on SIFT and GIST descriptors demonstrate that the approach achieves high accuracy and efficiency, outperforming three state‑of‑the‑art methods and scaling to two billion vectors.

Abstract

This paper introduces a product quantization-based approach for approximate nearest neighbor search. The idea is to decompose the space into a Cartesian product of low-dimensional subspaces and to quantize each subspace separately. A vector is represented by a short code composed of its subspace quantization indices. The euclidean distance between two vectors can be efficiently estimated from their codes. An asymmetric version increases precision, as it computes the approximate distance between a vector and a code. Experimental results show that our approach searches for nearest neighbors efficiently, in particular in combination with an inverted file system. Results for SIFT and GIST image descriptors show excellent search accuracy, outperforming three state-of-the-art approaches. The scalability of our approach is validated on a data set of two billion vectors.