Version: 1.4.0
Release date: 03-10-2017
Last Modified: 25-04-2018
This package is written in the C++ programming language. It contains two external LSH schemes QALSH and QALSH+ for high-dimensional c-Approximate Nearest Neighbor (or simply c-ANN) search under l_p norm, where p \in (0, 2].
We provide a Makefile and a shell script (i.e., run_mnist.sh) as a running example for comipling and running the package. Before running this package, please ensure the input format of the dataset and query set is correct. We give a sample dataset and query set (i.e., Mnist) for your reference.
We also share the shell scripts (i.e., run_sift.sh, run_gist.sh, run_trevi.sh, and run_p53.sh) for the users who would like to reproduce the results presented in PVLDB 2015 and VLDBJ 2017. The datasets Sift, Gist, Trevi, and P53 we used can be downloaded from the following links:
-
Sift: https://drive.google.com/open?id=1Q3_dnblolD9GVis7OakP2mrqmBApytEL
-
Gist: https://drive.google.com/open?id=1M3hJl5slY_pu50IQ7ie-t9E6RvzMizYT
-
Trevi: https://drive.google.com/open?id=1RF1FJKWHv3y7W7aBrewnOMrWR15dNbJ3
-
P53: https://drive.google.com/open?id=15mzraPmxNRzcfhXsd_KWBgKclUFUZQEj
-
Qiang Huang
Smart Systems Institute, National University of Singapore (NUS),
Singapore, 119613
-
Jianlin Feng
School of Data and Computer Science, Sun Yat-Sen University (SYSU),
Guangzhou, China, 510006
fengjlin@mail.sysu.edu.cn, fengjl9@gmail.com
http://sdcs.sysu.edu.cn/content/2511, http://ss.sysu.edu.cn/~fjl/
The papers for the package of QALSH have been published in PVLDB 2015 and VLDBJ 2017, which are displayed as follows:
-
Qiang Huang, Jianlin Feng, Yikai Zhang, Qiong Fang, and Wilfred Ng. Query-Aware Locality-Sensitive Hashing for Approximate Nearest Neighbor Search. In Proceedings of the VLDB Endowment (PVLDB), 9(1): 1 - 12, 2015.
-
Qiang Huang, Jianlin Feng, Qiong Fang, Wilfred Ng, and Wei Wang. Query-Aware Locality-Sensitive Hashing Scheme for l_p Norm. The VLDB Journal, 26(5): 683 – 708, 2017.
If you use the package for publications, please cite the papers above. Thank you.