Meta FAIR (FAISS) · Clustering Algorithm

e-means

e-means is a mini-batch k-means clustering algorithm from Meta FAIR that updates centroids using exponential moving averages, as a drop-in replacement for Lloyd's k-means. The authors report reaching Lloyd's 25-epoch loss in 1.4 to 2.6 epochs across 24 dataset configurations, with 10-45x GPU and 8-15x CPU speedups (self-reported). It ships in FAISS 1.16.0 under faiss.contrib.e_means and is MIT licensed.