Data clustering is a fundamental operation in data analysis. For handling large-scale data, the standard k-means clustering method is not only slow, but also memory-inefficient. We propose an efficient clustering method for billion-scale feature vectors, called PQk-means. By first compressing input vectors into short product-quantized (PQ) codes, PQk-means achieves fast and memory-efficient clustering, even for high-dimensional vectors. Similar to k-means, PQk-means repeats the assignment and update steps, both of which can be performed in the PQ-code domain. Experimental results show that even short-length (32 bit) PQ-codes can produce competitive results compared with k-means. This result is of practical importance for clustering in memory-restricted environments. Using the proposed PQk-means scheme, the clustering of one billion 128D SIFT features with K = 10^5 is achieved within 14 hours, using just 32 GB of memory consumption on a single computer.
@inproceedings{acmmm_matsui_2017,
author = {Yusuke Matsui and Keisuke Ogaki and Toshihiko Yamasaki and Kiyoharu Aizawa},
booktitle = {ACM International Conference on Multimedia (ACMMM)},
title = {PQk-means: Billion-scale Clustering for Product-quantized Codes},
pages = {1725--1733},
year = {2017}
}