Speeding up search with locality sensitive hashing

Locality sensitive hashing (LSH) is a technique for reducing complex data down to a simple hash code. If two hash codes are similar than the original data is similar. Typically, they are used for speeding up search and other similarity comparisons.

In this presentation I will discuss two ways of implementing LSH in python; the first method is completely stateless but only works on certain forms of data; the second is stateful but does not make any assumptions about the distribution of the underlying data. I will conclude the presentation by describing how we apply LSH to search at Lyst.