distributed_clustering

distributed_clustering(graph: TGraph, beta, rounds=None, patience=3, min_samples=2)

Distributed clustering algorithm

Implements algorithm of [1] with gpu support

Parameters:

• graph – input graph

• beta – \(\beta\) value of the algorithm (controls the number of seeds)

• rounds – number of iteration rounds (default: 3*int(log(graph.num_nodes)))

• patience – number of rounds without label changes before early stopping (default: 3)

• min_samples – minimum number of seed nodes (default: 2)

Reference

[1]

H. Sun and L. Zanetti. “Distributed Graph Clustering and Sparsification”. ACM Transactions on Parallel Computing 6.3 (2019), pp. 1–23. doi: 10.1145/3364208.