What is the LSH Forest algorithm?

To give our algorithm a better chance at finding out more close points, we could grow a forest of independent trees with their own random planes and return the top K points obtained from all of the trees. This algorithm is called as the LSH-forest algorithm. read more

This algorithm is called as the LSH-forest algorithm. You can also read a more detailed but probably easier to follow description of the algorithm in Erik Bernhardsson’ s excellent blog post : Nearest neighbors and vector models - part 2 - algorithms and data structures. read more

We analyze LSH Forest [BCG05]|a popular heuristic for the nearest neighbor search|and show that a careful yet simple modi cation of it outperforms \vanilla" LSH algorithms. The end result is the rst instance of a simple, practical algorithm that provably leverages data-dependent hashing to improve upon data-oblivious LSH. read more