Hierarchical Navigable Small Worlds
In this tutorial, we'll build on that knowledge by looking at today's most commonly used primary algorithm: Hierarchical Navigable Small Worlds (HNSW). HNSW performs very well regarding speed and accuracy, making it an incredibly robust vector search algorithm. Unfortunately, despite being popular, understanding HNSW can be tricky, but don't fret - in the next couple of sections, we'll break down HNSW into its steps, developing our own simple implementation along the way.
A follow-up to this piece on scalar quantization and a predecessor to Approximate Nearest Neighbors Oh Yeah. The series is [indexed here](from a series indexed here)