AI Search with HNSW: Hierarchical Navigable Small World (HNSW)

🔍 Hierarchical Navigable Small Worlds (HNSW): from construction to search

🚀 This is the last episode on Hierarchical Navigable Small Worlds (HNSW) - one of the top-performing approximate nearest neighbour search algorithms in vector databases. Here, we explore what HNSW is and how it works.

💡 HNSW is a multi-layered graph of hierarchical layers. It combines the best of probability skip lists (episode 6) and Navigable Small Worlds (episode 7), enabling robust and fast nearest-neighbor search.

🎥 In this episode, we explore:
■ Construction of HNSW
■ Approximate nearest neighbour search in HNSW

👍 Watch now and share your thoughts in the comments. If you find our videos useful, we appreciate a like

📚 Sources and Further Reading:
[1] Y. A. Malkov and D. A. Yashunin, IEEE Trans. Pattern. Anal. Mach. Intell. (2020), 42, 4, 824-836
[2] Y. Malkov et al., Approximate Nearest Neighbor Search Small World Approach (2011), International Conference on Information and Communication Technologies & Applications
[3] W. Pugh, Communications of the ACM (1990), 33, 6, 668-676.

#VectorDatabases #HNSW #ObjectBox #ANN #ArtificialIntelligence