Research Editor selected
Meta FAIR releases e-means clustering algorithm in FAISS 1.16.0
Meta FAIR released e-means, an EMA mini-batch drop-in replacement for Lloyd's k-means, shipping in FAISS 1.16.0 under faiss.contrib.e_means. The authors report it reaches Lloyd's 25-epoch loss in 1.4 to 2.6 epochs, with self-reported 10-45x GPU and 8-15x CPU speedups. It uses fitted scaling laws for batch size and step size, plus dead-cluster revival and fused GPU kernels. Tests span 100K to 10 billion points and up to 16.7M clusters.
Why it matters: Readers building vector indexes, quantizers or data-curation pipelines get a concrete, code-available clustering speedup with clear self-reported caveats.
Meta FAIRFAISSk-means clustering