cr.sparse.cluster.vq.kmeans_with_seed_jit¶
- cr.sparse.cluster.vq.kmeans_with_seed_jit(key, points, k, thresh=1e-05, max_iters=100)¶
Runs the k-means algorithm for a specific random initialization
- Parameters
key – a PRNG key used as the random key for choosing initial centroids
points (jax.numpy.ndarray) – Each row of the points matrix is a point.
k (int) – The number of clusters
thresh (float) – Convergence threshold on change in distortion
max_iters (int) – Maximum number of iterations for k-means algorithm
- Returns
A named tuple consisting of: centroids for each cluster, assignment of each point to a cluster, current distorition, previous distortion, number of iterations for convergence.
- Return type
