DmitryKey/big-ann

BigANN Experiments of Team Sisu

big-ann

Billion-Scale Approximate Nearest Neighbor Search Challenge: http://big-ann-benchmarks.com/index.html

The following blog describes BuddyPQ algorithm in detail, which achived 12% increase in recall over FAISS for 10M dataset:

Billion-Scale Vector Search: Team Sisu and BuddyPQ

Solutions implemented so far

1. Sharding
   1. Clustering K-Means based sharding algorithm
   2. KANNDI K Approximate Nearest Neighbours DIstance-based algorithm

Related projects

• Billion-Scale ANN Benchmarks: https://github.com/harsha-simhadri/big-ann-benchmarks
• Million-Scale ANN Benchmarks: https://github.com/erikbern/ann-benchmarks
• Getting practical with vector search in Solr and Elasticsearch: https://github.com/DmitryKey/bert-solr-search