ConcensusClust: clusters nodes by concensus (majority voting) initialized by...

In randnet: Random Network Model Estimation, Selection and Parameter Tuning

clusters nodes by concensus (majority voting) initialized by regularized spectral clustering

Description

community detection by concensus (majority voting) initialized by regularized spectral clustering

Usage

ConsensusClust(A,K,tau=0.25,lap=TRUE)

Arguments

A	adjacency matrix
K	number of communities
tau	reguarlization parameter for regularized spectral clustering. Default value is 0.25. Typically set between 0 and 1. If tau=0, no regularization is applied.
lap	indicator. If TRUE, the Laplacian matrix for initializing clustering. If FALSE, the adjacency matrix will be used.

Details

Community detection algorithm by majority voting algorithm of Gao et. al. (2016). When initialized by regularized spectral clustering, it is shown that the clustering accuracy of this algorithm gives minimax rate under the SBM. However, it can slow compared with spectral clustering.

Value

cluster labels

Author(s)

Tianxi Li, Elizaveta Levina, Ji Zhu

Maintainer: Tianxi Li <[e-mail]>

References

Gao, C.; Ma, Z.; Zhang, A. Y. & Zhou, H. H. Achieving optimal misclassification proportion in stochastic block models The Journal of Machine Learning Research, JMLR. org, 2017, 18, 1980-2024

See Also

reg.SP

Examples

dt <- BlockModel.Gen(15,50,K=2,beta=0.2,rho=0)


A <- dt$A


cc <- ConsensusClust(A,K=2,lap=TRUE)

randnet documentation built on May 31, 2023, 6:44 p.m.