Home

Documentation

FALCONN is a C++ library with a Python wrapper for similarity search over high-dimensional data. This page contains links to several documentation files.

We currently support GNU/Linux and OS X, but not Windows. In principle, one can run FALCONN under Cygwin, but the performance will be sub-optimal.

How to use FALCONN

If you are already familiar with LSH, you might just want to download FALCONN and get started.

• First, here are our installation instructions.
• Once you have set up FALCONN, you can read how to use FALCONN.
• We have a C++ API reference generated by Doxygen and a Python API reference generated by pdoc.

Background

In order to make best use of FALCONN, it is helpful to know about Locality-Sensitive Hashing (LSH).

• We have written a gentle introduction to LSH.
• Brief description of the hash families provided by FALCONN.

Further documentation

• We have collected a few FAQs.
• We provide an example for how to use FALCONN in order to perform similarity search over the GloVe datasets. This example should be helpful in making sense of the parameters FALCONN uses.